heart disease research paper

Editor's Choice: The Latest in Cardiology in JAMA

Nay Aung, PhD; Hannah L. Nicholls, PhD; C. Anwar A. Chahal, PhD; et al

Original Investigation | August 28, 2024

Just Published

Lesion-Level Effects of LDL-C–Lowering Therapy in Patients With Acute Myocardial Infarction Flavio G. Biccirè, MD; et al. Original Investigation free access Flavio G. Biccirè, MD; et al. Invited Commentary
Prehospital Pulse-Dose Glucocorticoid in STEMI Jasmine Melissa Madsen, MD; et al. Original Investigation free access Jasmine Melissa Madsen, MD; et al.
Effects of Sacubitril/Valsartan on All-Cause Hospitalizations in Heart Failure Henri Lu, MD; et al. Brief Report free access Henri Lu, MD; et al.
What We Have Learned About Reducing Low-Density Lipoprotein Cholesterol and Coronary Plaques Steven E. Nissen, MD Invited Commentary free access Steven E. Nissen, MD
Genotype- vs Phenotype-Guided Approaches to Improve ATTR Detection Jessica A. Regan, MD; et al. Editorial has multimedia Jessica A. Regan, MD; et al.
Reporting Primary Language in Cardiology Clinical Trials Zara Latif, MD; et al. Viewpoint Zara Latif, MD; et al.
Tombstone Pattern Electrocardiogram in a Young Woman Yun Chen, MD; et al. JAMA Cardiology Clinical Challenge has active quiz Yun Chen, MD; et al.
Racial Disparities in Sports Cardiology Sheela Krishnan, MD; et al. Review has active quiz Sheela Krishnan, MD; et al.
Climate Change and Cardiovascular Health Dhruv S. Kazi, MD, MSc, MS; et al. Review has active quiz Dhruv S. Kazi, MD, MSc, MS; et al.
46,237 Views Sex Differences in Blood Pressure Trajectories Over the Life Course
7,828 Views Prehospital Pulse-Dose Glucocorticoid in STEMI
7,135 Views Low-Dose Oral Anticoagulation vs Dual Antiplatelet Therapy After LAAO
5,478 Views Effects of Sacubitril/Valsartan on All-Cause Hospitalizations in Heart Failure
4,686 Views Leadless Ultrasound-Based Cardiac Resynchronization System in Heart Failure
4,667 Views Lipoprotein(a) and Long-Term Plaque Progression, Low-Density Plaque, and Pericoronary Inflammation
4,548 Views Rates of Sudden Death After Myocardial Infarction—VALIANT and PARADISE-MI Trials
3,983 Views Verinurad Plus Allopurinol for HFpEF
3,786 Views Tombstone Pattern Electrocardiogram in a Young Woman
3,182 Views Redefining Cardiac Involvement in Systemic Immunoglobulin AL Amyloidosis
383 Citations Myocarditis Following Immunization With mRNA COVID-19 Vaccines in Members of the US Military
238 Citations Patients With Acute Myocarditis Following mRNA COVID-19 Vaccination
156 Citations Clopidogrel Monotherapy After 1 to 2 Months of DAPT vs 12 Months of DAPT in Acute Coronary Syndrome
136 Citations TET2 -Driven Clonal Hematopoiesis and Response to Canakinumab
131 Citations Association of Myocarditis With BNT162b2 Vaccination in Children
122 Citations Prevalence of Coronary Artery Disease and Coronary Microvascular Dysfunction in Heart Failure
118 Citations SARS-CoV-2 Vaccination and Myocarditis in a Nordic Cohort Study of 23 Million Residents
115 Citations ApoB-Containing Lipoproteins and Risk of MI in Individuals With and Without Atherosclerosis
99 Citations Treatment With Sacubitril/Valsartan in Patients With Advanced Heart Failure and Reduced Ejection Fraction
96 Citations Lipoprotein(a) and its Significance in Cardiovascular Disease
Register for email alerts with links to free full-text articles
Access PDFs of free articles
Manage your interests
Save searches and receive search alerts

An official website of the United States government

The .gov means it’s official. Federal government websites often end in .gov or .mil. Before sharing sensitive information, make sure you’re on a federal government site.

The site is secure. The https:// ensures that you are connecting to the official website and that any information you provide is encrypted and transmitted securely.

Publications
Account settings
My Bibliography
Collections
Citation manager

Save citation to file

Email citation, add to collections.

Create a new collection
Add to an existing collection

Add to My Bibliography

Your saved search, create a file for external citation management software, your rss feed.

Search in PubMed
Search in NLM Catalog
Add to Search

A Brief Review of Cardiovascular Diseases, Associated Risk Factors and Current Treatment Regimes

Affiliation.

1 Department of Internal Medicine, University of Iowa, Iowa City, IA 52242, United States.
PMID: 31553287
DOI: 10.2174/1381612825666190925163827

Cardiovascular diseases (CVDs) are the leading cause of premature death and disability in humans and their incidence is on the rise globally. Given their substantial contribution towards the escalating costs of health care, CVDs also generate a high socio-economic burden in the general population. The underlying pathogenesis and progression associated with nearly all CVDs are predominantly of atherosclerotic origin that leads to the development of coronary artery disease, cerebrovascular disease, venous thromboembolism and, peripheral vascular disease, subsequently causing myocardial infarction, cardiac arrhythmias or stroke. The aetiological risk factors leading to the onset of CVDs are well recognized and include hyperlipidaemia, hypertension, diabetes, obesity, smoking and, lack of physical activity. They collectively represent more than 90% of the CVD risks in all epidemiological studies. Despite high fatality rate of CVDs, the identification and careful prevention of the underlying risk factors can significantly reduce the global epidemic of CVDs. Beside making favorable lifestyle modifications, primary regimes for the prevention and treatment of CVDs include lipid-lowering drugs, antihypertensives, antiplatelet and anticoagulation therapies. Despite their effectiveness, significant gaps in the treatment of CVDs remain. In this review, we discuss the epidemiology and pathology of the major CVDs that are prevalent globally. We also determine the contribution of well-recognized risk factors towards the development of CVDs and the prevention strategies. In the end, therapies for the control and treatment of CVDs are discussed.

Keywords: Atherosclerosis; epidemiological studies; hypertension; platelets; stroke; thrombosis..

PubMed Disclaimer

Publication types

Search in MeSH

Related information

Linkout - more resources, full text sources.

Bentham Science Publishers Ltd.
Ingenta plc
Citation Manager

NCBI Literature Resources

MeSH PMC Bookshelf Disclaimer

The PubMed wordmark and PubMed logo are registered trademarks of the U.S. Department of Health and Human Services (HHS). Unauthorized use of these marks is strictly prohibited.

An official website of the United States government

Here’s how you know

Official websites use .gov A .gov website belongs to an official government organization in the United States.

Secure .gov websites use HTTPS A lock ( A locked padlock ) or https:// means you’ve safely connected to the .gov website. Share sensitive information only on official, secure websites.

Heart-Healthy Living
High Blood Pressure
Sickle Cell Disease
Sleep Apnea
Information & Resources on COVID-19
The Heart Truth®
Learn More Breathe Better®
Blood Diseases & Disorders Education Program
Publications and Resources
Clinical Trials
Blood Disorders and Blood Safety
Sleep Science and Sleep Disorders
Lung Diseases
Health Disparities and Inequities
Heart and Vascular Diseases
Precision Medicine Activities
Obesity, Nutrition, and Physical Activity
Population and Epidemiology Studies
Women’s Health
Research Topics
All Science A-Z
Grants and Training Home
Policies and Guidelines
Funding Opportunities and Contacts
Training and Career Development
Email Alerts
NHLBI in the Press
Research Features
Ask a Scientist
Past Events
Upcoming Events
Mission and Strategic Vision
Divisions, Offices and Centers
Advisory Committees
Budget and Legislative Information
Jobs and Working at the NHLBI
Contact and FAQs
NIH Sleep Research Plan
< Back To Research Topics

Coronary Heart Disease Research

Language switcher.

For almost 75 years, the NHLBI has been at the forefront of improving the nation’s health and reducing the burden of heart and vascular diseases . Heart disease, including coronary heart disease, remains the leading cause of death in the United States. However, the rate of heart disease deaths has declined by 70% over the past 50 years, thanks in part to NHLBI-funded research. Many current studies funded by the NHLBI focus on discovering genetic associations and finding new ways to prevent and treat the onset of coronary heart disease and associated medical conditions.

NHLBI research that really made a difference

The NHLBI supports a wide range of long-term studies to understand the risk factors of coronary heart disease. These ongoing studies, among others, have led to many discoveries that have increased our understanding of the causes of cardiovascular disease among different populations, helping to shape evidence-based clinical practice guidelines.

Risk factors that can be changed: The NHLBI Framingham Heart Study (FHS) revealed that cardiovascular disease is caused by modifiable risk factors such as smoking, high blood pressure , obesity , high cholesterol levels, and physical inactivity. It is why, in routine physicals, healthcare providers check for high blood pressure, high cholesterol, unhealthy eating patterns, smoking, physical inactivity, and unhealthy weight. The FHS found that cigarette smoking increases the risk of heart disease. Researchers also showed that cardiovascular disease can affect people differently depending on sex or race, underscoring the need to address health disparities.
Risk factors for Hispanic/Latino adults: The Hispanic Community Health Study/Study of Latinos (HCHS/SOL) found that heart disease risk factors are widespread among Hispanic/Latino adults in the United States , with 80% of men and 71% of women having at least one risk factor. Researchers also used HCHS/SOL genetic data to explore genes linked with central adiposity (the tendency to have excess body fat around the waist) in Hispanic/Latino adults. Before this study, genes linked with central adiposity, a risk factor for coronary heart disease, had been identified in people of European ancestry. These results showed that those genes also predict central adiposity for Hispanic/Latino communities. Some of the genes identified were more common among people with Mexican or Central/South American ancestry, while others were more common among people of Caribbean ancestry.
Risk factors for African Americans: The Jackson Heart Study (JHS) began in 1997 and includes more than 5,300 African American men and women in Jackson, Mississippi. It has studied genetic and environmental factors that raise the risk of heart problems, especially high blood pressure, coronary heart disease, heart failure , stroke , and peripheral artery disease (PAD) . Researchers discovered a gene variant in African American individuals that doubles the risk of heart disease. They also found that even small spikes in blood pressure can lead to a higher risk of death. A community engagement component of the JHS is putting 20 years of the study’s findings into action by turning traditional gathering places, such as barbershops and churches, into health information hubs.
Risk factors for American Indians: The NHLBI actively supports the Strong Heart Study , a long-term study that began in 1988 to examine cardiovascular disease and its risk factors among American Indian men and women. The Strong Heart Study is one of the largest epidemiological studies of American Indian people ever undertaken. It involves a partnership with 12 Tribal Nations and has followed more than 8,000 participants, many of whom live in low-income rural areas of Arizona, Oklahoma, and the Dakotas. Cardiovascular disease remains the leading cause of death for American Indian people. Yet the prevalence and severity of cardiovascular disease among American Indian people has been challenging to study because of the small sizes of the communities, as well as the relatively young age, cultural diversity, and wide geographic distribution of the population. In 2019, the NHLBI renewed its commitment to the Strong Heart Study with a new study phase that includes more funding for community-driven pilot projects and a continued emphasis on training and development. Read more about the goals and key findings of the Strong Heart Study.

Current research funded by the NHLBI

Within our Division of Cardiovascular Sciences , the Atherothrombosis and Coronary Artery Disease Branch of its Adult and Pediatric Cardiac Research Program and the Center for Translation Research and Implementation Science oversee much of our funded research on coronary heart disease.

Research funding

Find funding opportunities and program contacts for research on coronary heart disease.

Current research on preventing coronary heart disease

Blood cholesterol and coronary heart disease: The NHLBI supports new research into lowering the risk of coronary heart disease by reducing levels of cholesterol in the blood. High levels of blood cholesterol, especially a type called low-density lipoprotein (LDL) cholesterol, raise the risk of coronary heart disease. However, even with medicine that lowers LDL cholesterol, there is still a risk of coronary heart disease due to other proteins, called triglyceride-rich ApoB-containing lipoproteins (ApoBCLs), that circulate in the blood. Researchers are working to find innovative ways to reduce the levels of ApoBCLs, which may help prevent coronary heart disease and other cardiovascular conditions.
Pregnancy, preeclampsia, and coronary heart disease risk: NHLBI-supported researchers are investigating the link between developing preeclampsia during pregnancy and an increased risk for heart disease over the lifespan . This project uses “omics” data – such as genomics, proteomics, and other research areas – from three different cohorts of women to define and assess preeclampsia biomarkers associated with cardiovascular health outcomes. Researchers have determined that high blood pressure during pregnancy and low birth weight are predictors of atherosclerotic cardiovascular disease in women . Ultimately, these findings can inform new preventive strategies to lower the risk of coronary heart disease.
Community-level efforts to lower heart disease risk among African American people: The NHLBI is funding initiatives to partner with churches in order to engage with African American communities and lower disparities in heart health . Studies have found that church-led interventions reduce risk factors for coronary heart disease and other cardiovascular conditions. NHLBI-supported researchers assessed data from more than 17,000 participants across multiple studies and determined that these community-based approaches are effective in lowering heart disease risk factors .

Find more NHLBI-funded studies on preventing coronary heart disease on the NIH RePORTER.

Learn about the impact of COVID-19 on your risk of coronary heart disease.

Current research on understanding the causes of coronary heart disease

Pregnancy and long-term heart disease: NHLBI researchers are continuing the Nulliparous Pregnancy Outcomes Study: Monitoring Mothers-to-be (nuMoM2b) study to understand the relationship between pregnancy-related problems, such as gestational hypertension, and heart problems. The study also looks at how problems during pregnancy may increase risk factors for heart disease later in life. NuMoM2b launched in 2010, and long-term studies are ongoing, with the goal of collecting high-quality data and understanding how heart disease develops in women after pregnancy.
How coronary artery disease affects heart attack risk: NHLBI-funded researchers are investigating why some people with coronary artery disease are more at risk for heart attacks than others. Researchers have found that people with coronary artery disease who have high-risk coronary plaques are more likely to have serious cardiac events, including heart attacks. However, we do not know why some people develop high-risk coronary plaques and others do not. Researchers hope that this study will help providers better identify which people are most at risk of heart attacks before they occur.
Genetics of coronary heart disease: The NHLBI supports studies to identify genetic variants associated with coronary heart disease . Researchers are investigating how genes affect important molecular cascades involved in the development of coronary heart disease . This deeper understanding of the underlying causes for plaque buildup and damage to the blood vessels can inform prevention strategies and help healthcare providers develop personalized treatment for people with coronary heart disease caused by specific genetic mutations.

Find more NHLBI-funded studies on understanding the causes of coronary heart disease on the NIH RePORTER.

Recent findings suggest that cholesterol-lowering treatment can lower the risk of heart disease complications in people with HIV.

Current research on treatments for coronary heart disease

Insight into new molecular targets for treatment: NHLBI-supported researchers are investigating the role of high-density lipoprotein (HDL) cholesterol in coronary heart disease and other medical conditions . Understanding how the molecular pathways of cholesterol affect the disease mechanism for atherosclerosis and plaque buildup in the blood vessels of the heart can lead to new therapeutic approaches for the treatment of coronary heart disease. Researchers have found evidence that treatments that boost HDL function can lower systemic inflammation and slow down plaque buildup . This mechanism could be targeted to develop a new treatment approach for coronary heart disease.
Long-term studies of treatment effectiveness: The NHLBI is supporting the International Study of Comparative Health Effectiveness with Medical and Invasive Approaches (ISCHEMIA) trial EXTENDed Follow-up (EXTEND) , which compares the long-term outcomes of an initial invasive versus conservative strategy for more than 5,000 surviving participants of the original ISCHEMIA trial. Researchers have found no difference in mortality outcomes between invasive and conservative management strategies for patients with chronic coronary heart disease after more than 3 years. They will continue to follow up with participants for up to 10 years. Researchers are also assessing the impact of nonfatal events on long-term heart disease and mortality. A more accurate heart disease risk score will be constructed to help healthcare providers deliver more precise care for their patients.
Evaluating a new therapy for protecting new mothers: The NHLBI is supporting the Randomized Evaluation of Bromocriptine In Myocardial Recovery Therapy for Peripartum Cardiomyopathy (REBIRTH) , for determining the role of bromocriptine as a treatment for peripartum cardiomyopathy (PPCM). Previous research suggests that prolactin, a hormone that stimulates the production of milk for breastfeeding, may contribute to the development of cardiomyopathy late in pregnancy or the first several months postpartum. Bromocriptine, once commonly used in the United States to stop milk production, has shown promising results in studies conducted in South Africa and Germany. Researchers will enroll approximately 200 women across North America who have been diagnosed with PPCM and assess their heart function after 6 months.
Impact of mental health on response to treatment: NHLBI-supported researchers are investigating how mental health conditions can affect treatment effectiveness for people with coronary heart disease. Studies show that depression is linked to a higher risk for negative outcomes from coronary heart disease. Researchers found that having depression is associated with poor adherence to medical treatment for coronary heart disease . This means that people with depression are less likely to follow through with their heart disease treatment plans, possibly contributing to their chances of experiencing worse outcomes. Researchers are also studying new ways to treat depression in patients with coronary heart disease .

Find more NHLBI-funded studies on treating coronary heart disease on the NIH RePORTER.

Researchers have found no clear difference in patient survival or heart attack risk between managing heart disease through medication and lifestyle changes compared with invasive procedures.

Coronary heart disease research labs at the NHLBI

Laboratory of Cardiac Physiology
Laboratory of Cardiovascular Biology
Minority Health and Health Disparities Population Laboratory
Social Determinants of Obesity and Cardiovascular Risk Laboratory
Laboratory for Cardiovascular Epidemiology and Genomics
Laboratory for Hemostasis and Platelet Biology

Related coronary heart disease programs

In 2002, the NHLBI launched The Heart Truth® , the first federally sponsored national health education program designed to raise awareness about heart disease as the leading cause of death in women. The NHLBI and The Heart Truth® supported the creation of the Red Dress® as the national symbol for awareness about women and heart disease, and also coordinate National Wear Red Day ® and American Heart Month each February.
The Biologic Specimen and Data Repository Information Coordinating Center (BioLINCC) facilitates access to and maximizes the scientific value of NHLBI biospecimen and data collections. A main goal is to promote the use of these scientific resources by the broader research community. BioLINCC serves to coordinate searches across data and biospecimen collections and provide an electronic means for requesting additional information and submitting requests for collections. Researchers wanting to submit biospecimen collections to the NHLBI Biorepository to share with qualified investigators may also use the website to initiate the application process.
Our Trans-Omics for Precision Medicine (TOPMed) Program studies the ways genetic information, along with information about health status, lifestyle, and the environment, can be used to predict the best ways to prevent and treat heart, lung, blood, and sleep disorders. TOPMed specifically supports NHLBI’s Precision Medicine Activities.
NHLBI population and epidemiology studies in different groups of people, including the Atherosclerosis Risk in Communities (ARIC) Study , the Multi-Ethnic Study of Atherosclerosis (MESA) , and the Cardiovascular Health Study (CHS) , have made major contributions to understanding the causes and prevention of heart and vascular diseases, including coronary heart disease.
The Cardiothoracic Surgical Trials Network (CTSN) is an international clinical research enterprise that studies heart valve disease , arrhythmias , heart failure, coronary heart disease, and surgical complications. The trials span all phases of development, from early translation to completion, and have more than 14,000 participants. The trials include six completed randomized clinical trials, three large observational studies, and many other smaller studies.

The Truth About Women and Heart Disease Fact Sheet

Learn how heart disease may be different for women than for men.

Explore more NHLBI research on coronary heart disease

The sections above provide you with the highlights of NHLBI-supported research on coronary heart disease. You can explore the full list of NHLBI-funded studies on the NIH RePORTER .

To find more studies:

Type your search words into the Quick Search box and press enter.
Check Active Projects if you want current research.
Select the Agencies arrow, then the NIH arrow, then check NHLBI .

If you want to sort the projects by budget size — from the biggest to the smallest — click on the FY Total Cost by IC column heading.

Research article
Open access
Published: 20 June 2018

Cardiovascular disease (CVD) and associated risk factors among older adults in six low-and middle-income countries: results from SAGE Wave 1

Ye Ruan 1 na1 ,
Yanfei Guo 1 na1 ,
Yang Zheng 1 ,
Zhezhou Huang 1 ,
Shuangyuan Sun 1 ,
Paul Kowal 2 , 3 ,
Yan Shi 1 &

BMC Public Health volume 18 , Article number: 778 ( 2018 ) Cite this article

26k Accesses

133 Citations

134 Altmetric

Metrics details

Cardiovascular disease (CVD) is one of the leading causes of death worldwide. Our study aimed to investigate the prevalence of two conditions, angina and stroke, and relevant risk factors among older adults in six low- and middle- income countries(LMICs).

The data was from World Health Organization (WHO) Study on global AGEing and adult Health (SAGE) Wave 1 in China, Ghana, India, Mexico, Russian Federation and South Africa. Presence of CVD was based on self-report of angina and stroke. Multivariate logistic regression was performed to examine the relationship between CVD and selected variables, including age, sex, urban/rural setting, household wealth, and risk factors such as smoking, alcohol drinking, fruit/vegetable intake, physical activity and BMI.

The age standardized prevalence of angina ranged from 9.5 % (South Africa) to 47.5 % (Russian Federation), and for stoke from 2.0% (India) to 6.1 % (Russia). Hypertension was associated with angina in China, India and Russian Federation after adjustment for age, sex, urban/rural setting, education and marital status (OR ranging from 1.3 [1.1-1.6] in India to 3.8 [2.9-5.0] in Russian Federation), furthermore it was a risk factor of stroke in five countries except Mexico. Low or moderate physical activity were also associated with angina in China, and were also strongly associated with stroke in all countries except Ghana and India. Obesity had a stronger association with angina in Russian Federation and China(ORs were 1.5[1.1-2.0] and 1.2 [1.0-1.5] respectively), and increased the risk of stroke in China. Smoking was associated with angina in India and South Africa(ORs were 1.6[1.0-2.4] and 2.1 [1.2-3.6] respectively ), and was also a risk factor of stroke in South Africa. We observed a stronger association between frequent heavy drinking and stroke in India. Household income was associated with reduced odds of angina in China, India and Russian Federation, however higher household income was a risk factor of angina in South Africa.

While the specific mix of risk factors contribute to disease prevalence in different ways in these six countries – they should all be targeted in multi-sectoral efforts to reduce the high burden of CVD in today’s society.

Peer Review reports

Cardiovascular diseases (CVDs) are by far the leading cause of death in the world. An estimated 17.9 million people died from CVDs in 2015. Ischemic heart disease (IHD) and stroke were the top two leading causes of CVD health lost in each world region [ 1 , 2 ]. By 2030 more than 22.2 million people will die annually from CVDs. Populations in low and middle income countries (LMICs) now contribute 75% of the CVD deaths, which leads to 7% reduction of gross domestic product(GDP) in these countries [ 3 ].

A larger proportion of the global burden of CVDs is now borne by LMICs than in high income countries, this is despite a comparatively lower burden from risk factors in low compared to high income countries [ 4 , 5 , 6 ]. Given the high prevalence of CVD among older adults in LMIC, the projected increases in this population will be a major challenge for the health care system. Twenty-three percent of the total global burden of disease(GBD) was attributed to disorders in people aged 60 years and older. The main contributors to disease burden were CVDs, accounting for 30.3% of the total burden in older people in 2010 [ 7 ]. Reliable and comparable analysis of risks to CVD is especially important for projecting future disease burden and for shaping disease prevention efforts.

A number of population-based studies from lower income countries have suggested that socio-demographic characteristics are associated with CVD, with increasing age, female sex and lower education consistently associated with higher prevalence of CVD. Some epidemiological evidence also suggests that CVD is associated with behavioral risk factors such as smoking, alcohol use, low physical activity levels, and insufficient vegetable and fruit intake, hypertension is also regarded as a very important risk factor for CVD. Independently or in combination, these risk factors present an opportunity for interventions to reduce future CVD burdens in ageing populations in LMIC.A number of large recent studies have compared CVD risks in higher and lower income countries, providing valuable and needed information about CVD and CVD risks [ 4 , 5 , 6 ]. However, the results of these studies may not be representative of the older adult population. For example, the Prospective Urban Rural Epidemiology (PURE) study sampling strategy and distributions provide less reliable estimates at older ages [ 8 ]. The World Health Organization Study on global AGEing and adult health (SAGE) is focused on older adults and use similar methodology across countries to improve comparability of important covariates and disease prevalence. Three of the countries overlap in PURE and SAGE (China, India and South Africa) where SAGE includes three additional middle income countries (Ghana, Mexico and the Russian Federation).

The aim of the present study was to investigate the prevalence of two main CVDs (angina, stroke) and behavioural risk factors and associated social-economic status (SES) factors among older adults using a unique data set with nationally representative samples in six low and middle income countries.

Sample and procedure

The data was from World Health Organization (WHO) Study on Global AGEing and adult health (SAGE) Wave 1, a longitudinal cohort study of ageing and older adults from 2007 to 2010 in six low- and middle-income countries (China, Ghana, India, Mexico, Russian Federation and South Africa) [ 9 ]. SAGE Wave 1 used face-to-face individual interviews to capture data. All six countries implemented multistage cluster sampling strategies which resulted in nationally representative cohorts of older adults ( http://www.who.int/healthinfo/sage/SAGEWorkingPaper5_Wave1Sampling.pdf?ua=1 ). Response rates for SAGE countries were Mexico 51%, India 68%, Ghana 80%, Russian Federation 83%, South Africa 77% and China 93%. Examination of non-respondent data suggested non-significant differences on some covariates (data not shown). Data were obtained following application for access through http://apps.who.int/healthinfo/systems/surveydata/index.php/catalog .

SAGE has been approved by the World Health Organization's Ethical Review Board. Additionally, each partner organization obtained ethical clearance through their respective review bodies. All study participants signed informed consent.

CVDs conditions

Two methods of assessing presence or absence of CVD were used. One was based on self-report of angina or stroke; and the second used an algorithm based on validated symptom-reporting methods to estimate and compare prevalence rates.

Sociodemographic variables

Socio-demographic variables contain age, sex, education, rural/urban residence, and income quintiles. Age was categorized into four groups: 50 to 59 years; 60 to 69 years; 70 to 79 years; and 80 years or older. Education level was classified into seven categories for analysis using an international classification scheme [ 10 ]. The income quintiles were generated using an asset-based approach- possession of assets and dwelling characteristics [ 11 ], with quintile 1(Q1) the quintile of the poorest households and quintile 5(Q5) the quintile of the richest.

Risk factors

Tobacco use.

Tobacco use was assessed by self-report and included different forms (manufactured or hand-rolled cigarettes, cigars, cheroots or whether tobacco is smoked, chewed, sucked or inhaled), and frequency of smoking, snuffing or chewing in each day over the week before interview[ 12 ], classified into four groups: never smoker, not current smokers, smokers(not daily) and current daily smokers.

Alcohol consumption

Alcohol consumption was categorized into four groups: life time abstainer, non-heavy drinkers, infrequent heavy drinkers and frequent heavy drinkers according to the consumption number of standard drinks of beer, wine and or spirit, fermented cider, and other alcoholic drinks during the week before interview.

Physical activity

Physical activity was measured by the Global Physical Activity Questionnaire (GPAQ) and assessed intensity, duration, and frequency of physical activity in three domains: occupational, transport-related, and discretionary or leisure time. Based on a standard classification scheme, three categories were generated: low, moderate and high levels [ 13 ].

Fruit and vegetable consumption

Fruit and vegetable consumption was assessed according to the number of daily servings eaten – with each serving approximating 80 grams. Five or more servings were defined as sufficient daily intake (at least 400 grams per day), fewer than five servings were categorized as insufficient [ 14 ].

Hypertension

The definition of hypertension used was systolic blood pressure ≥140mmHg and/or diastolic blood pressure ≥ 90mmHg and/or self-reported treatment with antihypertensive medication during the two weeks before interview. Blood pressure measurements were conducted three times on the right arm of the seated respondent with an automated recording device (OMRON R6 Wrist Blood Pressure Monitor, HEM-6000-E, Omron Healthcare Europe), and calculated as an average of the latter two measurements.

According to the classification criteria proposed by the WHO [ 15 ], body mass index (BMI) of <18.5 kg/m 2 , 25–29.9 kg/m 2 and ≥30 kg/m 2 are used to define underweight, overweight and obesity, respectively. Modified BMI cutoffs for China and India were used to perform an additional set of analyses that describes overweight (BMI 23.0-27.5) and obesity (BMI >27.5) in Asian populations [ 16 ].

Statistical methods

Statistic analyse were conducted using STATA SE version 11 (Stata Corp, College Station, TX). The prevalence of angina and stroke were calculated by using normalized weights in each country. Weights were based on selection probability, non-response, and post-stratification adjustments. To improve comparability across countries, the prevalence rates were age-standardized using the WHO World Standard Population Distribution based on world average population 2000-2025 [ 17 ]. Multivariate logistic regression was performed to examine the relationship between CVD and selected variables, including the socio-demographics such as age, sex, urban/rural setting, education, household wealth, and health risk factors such as smoking, alcohol drinking, fruit/vegetable intake, physical activity, hypertension and obesity. P < 0.05 from two-sided statistical tests was considered statistically significant.

A total of 34,114 individuals were included in the final analyses. Table 1 shows the sample distribution and demographic, socioeconomic and lifestyle characteristics by countries. The proportions of women are higher than men in four countries, except Ghana and India. The majority of older Indian lived in rural locations, while compared to urban areas in the other countries. The 50-59 age groups had the largest proportions in all countries, but the SAGE sample population distributions match those of the United Nations and US Census Bureau’s International Data Base estimates [ 18 ]. The percentage of respondents with no formal education were higher in Ghana (54.0%) and India (51.2%). In contrast, Russian Federation had the highest educational level with only 0.5% with no formal education and over 20% with a college degree or higher.

The rate of daily smoking ranged from 7.6% (Ghana) to 46.9% (India), frequent heavy drinker was the highest in China (6.4%) and lowest in Mexico (0.1%), and the highest rate of low physical activity was in South Africa (59.5%). Insufficient fruit and vegetable intake was more common in India, the Russian Federation and Mexico (90.6, 81.0 and 81.4%, respectively) compared with China, South Africa and Ghana (35.7, 68.5 and 68.9%, respectively).

The age standardized prevalence of angina ranged from 9.5 % (South Africa) to 47.5 % (Russian Federation). It was higher in women than in men in all six countries. The rates were higher in rural than in urban locations other than in China. Angina rose with age in each country except Mexico, and a slight drop was seen in the highest age group in Ghana, India, Russian Federation and South Africa. The lowest prevalence of angina was found in individuals with the highest household income in China, Ghana, India and Russian Federation, respectively (see Table 2 ).

The prevalence of stroke was 6.1% in Russian Federation, which was higher than the other SAGE countries, while India had the lowest prevalence of 2.0%. In Russian Federation, the prevalence of stroke in men was almost twice that of women. Stroke was higher in urban than in rural locations in all six countries. Stroke prevalence tended to increase with age in all SAGE countries, but a slight drop in 80+ age group in Mexico and Russian Federation. In China, the wealthiest older adults had the lowest stroke prevalence (see Table 3 ).

Table 4 shows the Odds ratios for likelihood of angina by risk factors. Hypertension was associated with angina in China, India and Russian Federation after adjustment for age, sex, urban/rural setting and education (OR ranging from 1.32 [1.13-1.55] in India to 3.80 [2.91-4.96] in Russian Federation). Low and moderate physical activity was also associated with angina in China (ORs were 1.46 [1.22-1.76] and 1.66[1.39-1.99], respectively). Obesity had a stronger association with angina in Russian Federation and China (ORs were 1.48[1.08-2.02] and 1.24[1.01-1.53], respectively). Smoking was associated with angina in India and South Africa (ORs were 1.56[1.02-2.36] and 2.11 [1.23-3.61], respectively). Non-heavy drinking was a protective factor for angina in China (OR was 0.67[0.51-0.87]). The OR (1.56[1.19-2.05]) for insufficient fruits and vegetables intake was highest in Ghana. Household income was associated with reduced odds ratios of angina in China, India and Russian Federation, however higher household income was a risk factor of angina in South Africa (see Table 4 ).

In all six LIMCs except Mexico, hypertension was associated with stroke (OR ranging from 1.98[1.04-3.80] in Ghana to 3.16[1.72-5.83] in Russian Federation). Low, moderate physical activity were also strongly associated with stroke in four LMICs apart from Ghana and India. In China, Obesity increased the risk of stroke (OR was 1.66[1.20-2.28]). Smoking was also a risk factor of stroke in South Africa. We observed a stronger association between frequent heavy drinking and stroke in India (OR 6.64[1.39 – 31.82]). Insufficient fruit and vegetable intake and household income were not significantly associated with stroke in any of the countries (see Table 5 ).

This study reports the prevalence of two common cardiovascular diseases, angina and stroke, and the relevant risk factors among older adults in six LIMCs. Globally, the age-adjusted CVDs mortality continues to be unevenly distributed: where it has decreased in high income countries(HICs) by 43% in recent decades [ 19 ], while LIMCs are drowning in a rising tide of CVD. Although age-standardized rates of death attributable to CVD declined 13% in LMICs from 381 per 100000 in 1990 to 332 per 100000 in 2013, the number of deaths increased 66% from 7.21 million to 12 million in 2013 with ageing and population growth ascribed as the main drivers [ 19 ]. Ischaemic heart disease and cerebrovascular disease (stroke) combined accounted for more than 85.1% of all cardiovascular disease deaths in 2016[ 20 ]. Our study indicated that CVDs(angina, stroke) were prevalent and variable among older adults in six countries. Angina and stroke were both highest in Russian Federation(47.5%, 6.1% respectively). Women were more likely to have angina than men in all six countries. Stroke was more prevalent in urban than in rural. Angina and stroke both tended to increase with age in China.

Prevalence of CVDs generally appeared to be most closely linked to a country’s stage of epidemiological transition [ 21 ], especially when high disease rates in middle age carry through into older ages. Underlying social, environmental, and economic shifts in many countries have led to increasing levels of predominant causes such as tobacco and alcohol use, sedentary lifestyle, unhealthy diets, and suboptimum levels of weight, blood pressure, cholesterol, and plasma glucose. The high and growing prevalence of CVD in LIMCs largely reflects the burden of these key risk factors. Our study revealed that hypertension, high BMI, decreased physical activity, frequent heavy drinking and lower household health were key risk factors of angina and stroke. However, the distribution of risk factors in six counties was unequal, for example, the factor with highest OR of angina in China and Russian Federation was hypertension, whereas it was smoking in India and South Africa.

Hypertension has been shown to be an independent risk factor for acute myocardial infarction and stroke in older people [ 22 , 23 ]. We found that hypertension was associated with angina in China, India and Russian Federation, in addition it was a risk factor of stroke in five of the six countries in this study (not Mexico). Between 1980 and 2008, blood pressure decreased by 2.0mmHg or more (for men) and 3.5 mmHg or more (for women) per decade in western Europe and Australia but increased by up to 2.7 mmHg over this same period in Oceania, East and West Africa and South and Southeast Asia [ 24 ]. Systematic review revealed that blood pressure lowering greatly reduced the major cardiovascular disease events and all-cause mortality, irrespective of starting blood pressure [ 25 ].However among these six LIMCs 66% hypertensives were undiagnosed before the survey, 73% untreated and 90% uncontrolled. Although the proportions of undiagnosed and untreated were lowest in Russia (30% and 35%), the uncontrolled rate was higher (83%) [ 26 ], low level of health care (primary and secondary prevention) and irregular treatment continued to be a major problem [ 27 ]. Hence, further research on early screening strategies, available health care and effective treatment of hypertension may be critical for improving outcomes.

Our study also showed that low physical activity and obesity besides hypertension were both associated with angina and stroke in China, and insufficient fruit and vegetable intake was risk factor of angina in Ghana. Compared with data from 1997, total physical activity in 2009 has decreased by 29% in males and by 38% in females in China [ 28 ], and physical inactivity was estimated the third leading risk factor for coronary heart disease [ 29 ]. As the relation between physical and obesity well recognized, obesity was an important risk factor of CVD. People are becoming more and more obese. Global age-standardised mean BMI increased from 21.7 kg/m 2 in 1975 to 24.2 kg/m 2 in 2014 in men, and from 22.1 kg/m 2 in 1975 to 24.4 kg/m 2 in 2014 in women. Over this period, age-standardised prevalence of obesity increased from 3.2% in 1975 to 10.8% in 2014 in men, and from 6·4 to 14.9% in women [ 30 ]. More than 50% of the obese individuals in the world lived in just 10 countries (listed in order of number of obese individuals): USA, China, India, Russia, Brazil, Mexico, Egypt, Pakistan, Indonesia, and Germany, and China and India jointly accounting for 15% in 2013[ 31 ]. China has moved from 60th place for men and 41st place for women in 1975 to second for both men and women in 2014 in the worldwide ranking of the number of severely obese individuals [ 30 ]. Unfortunately, the prevalence of obesity among children and adolescents are both on the rise. In comparison with obesity rate in 1985, it increased by 8.7 times for children and 38.1 times for adolescents [ 32 ]. In the World Health Survey 2002-2003, prevalence of low fruits and vegetable consumption among individuals aged 18-99 years in Ghana was the lowest among 52 countries [ 33 ]. However, the prevalence was higher (68.9%) among persons aged 50 years and older [ 34 ]. We also found that insufficient fruits and vegetables intake was associated with angina in Ghana. All of these contribute to the increasing burden of CVD.

We observed a relationship between smoking and angina, frequent heavy drinking and stroke in India. The prevalence of angina was 19.6% (95%CI:16.5-23.0) in India, the second highest for these six countries. CVD-related conditions contributed nearly two-thirds of the burden of NCD mortality in India [ 35 ], with ischemic heart disease(IHD) and stroke contributing substantially to CVD mortality in India (83%) [ 36 ].Up to 35% of adults in India consume tobacco [ 37 ], with the rate of daily tobacco use was highest(46.9%) among the six LIMCs in this study, highest in younger individuals (20–35 years) [ 38 ].The relation between alcohol consumption and CVD has been widely studied. Several analyses showed that low-moderate levels of alcohol consumption had cardio protective effects, while heavy drinking is harmful, usually described as “U-shaped” or “J-shaped” relationship [ 39 , 40 ]. Aside from alcohol consumption, drinking pattern (binge-pattern drinking) played an important role in elevating the risk of CVD [ 41 , 42 ]. Another cohort study showed that heterogeneous associations exist between level of alcohol consumption and CVD: compared with moderate drinking, heavy drinking raised risk of coronary death, heart failure, cardiac arrest, ischaemic stroke but a lower risk of myocardial infarction or stable angina [ 43 ]. We found that in China non-heavy drinking was a protective factor for angina and stroke, and frequent heavy drinking showed a dangerous effect for stroke in India.

There were a few limitations in our study. Firstly, although SAGE assembled nationally representative cohorts from six countries, the response rates were different across the countries, ranging from 51% in Mexico to 93% in China. The low response rate in Mexico was for specific reasons related to timing of the survey and inability to engage in repeat visits to households to maintain the sample and we note this introduces the potential for selection bias into the results for Mexico. Secondly, the data for stroke and some risk factors were based on self-reports, which may lead to recall bias. However, validated symptom-reporting methods were also used in these analyses to estimate and compare prevalence rates for angina to improve prevalence estimates. Thirdly, the question on stroke in SAGE did not distinguish between ischemic stroke and hemorrhagic stroke. Last, these results are based on cross-sectional data and as such, cannot be sure of the direction of the associations we identified.

Conclusions

In conclusion, our study provided representative prevalence of angina and stroke and relevant risk factors in elders in six LIMCs. Due to the variation pattern of prevalence and risk factors distribution, policies and health interventions will need to be targeted and tailored for a broad range of local conditions to achieve the health goals set by the United Nations for 2025.

Abbreviations

Body mass index

Cardiovascular disease

Global burden of disease

Gross national income

Global Physical Activity Questionnaire

High income countries

Ischemic heart disease

Low- and middle- income countries

Study on global AGEing and adult Health

Social-economic status

World Health Organization

Roth GA, Johnson C, Abajobir A, Abd-Allah F, Abera SF, Abyu G, et al. Global regional, and national burden of cardiovascular diseases for 10 causes, 1990 to 2015. J Am College Cardiol. 2017;70(1):1–25.

Article Google Scholar

Cardiovascular diseases (CVDs) Fact sheet. Available at: http://www.who.int/en/news-room/fact-sheets/detail/cardiovascular-diseases-(cvds) . Accessed 15 Mar 2017.

World Health Organization. Burden: mortality, morbidity and risk factors. In: Alwan A, editor. Global Status Report on Noncommunicable Diseases 2010. Geneva: World Health Organization; 2011.

Gaziano TA, Bitton A, Anand S, Abrahams-Gessel S, Murphy A. Growing Epidemic of Coronary Heart Disease in Low- and Middle-Income Countries. Curr Probl Cardiol. 2010;35:72–115.

Article PubMed PubMed Central Google Scholar

Yusuf S, Rangarajan S, Teo K, Islam S, Li W, Liu L, et al. Cardiovascular risk and events in 17 low-, middle-, and high-income countries. N Engl J Med. 2014;371:818–27.

Article PubMed CAS Google Scholar

Teo KK, Dokainish H. The Emerging Epidemic of Cardiovascular Risk Factors and Atherosclerotic Disease in Developing Countries. Can J Cardiol. 2017;33:358–65.

Article PubMed Google Scholar

Prince MJ, Wu F, Guo Y, Gutierrez Robledo LM, O'Donnell M, Sullivan R, et al. The burden of disease in older people and implications for health policy and practice. Lancet. 2015;385:549–62.

Corsi DJ, Subramanian SV, Chow CK, et al. Prospective Urban Rural Epidemiology (PURE) study: Baseline characteristics of the household sample and comparative analyses with national data in 17 countries. American Heart Journal. 2013;166:636–46.

Kowal P, Chatterji S, Naidoo N, Biritwum R, Fan W, Lopez Ridaura R, et al. Data resource profile: the World Health Organization Study on global AGEing and adult health (SAGE). Int J Epidemiol. 2012;41:1639–49.

United Nations Educational Scientific and Cultural Organization. International Standard Classification of Education (ISCED). 1997. ISBN 92-9189-035-9. Available at: http://uis.unesco.org/sites/default/files/documents/international-standard-classification-of-education-1997-en_0.pdf .

Ferguson BD, Tandon A, Gakidou E, Murray CJL. Estimating permanent income using asset and indicator variables. In: Evans DE, Murray CJL, editors. Health systems performance assessment debates, methods and empiricism. Geneva: World Health Organization; 2003. p. 747–60.

Google Scholar

World Health Organization. Guidelines for controlling and monitoring the tobacco epidemic. Geneva: World Health Organization; 1998. p. 76–101.

World Health Organization. Global Physical Activity Questionnaire (GPAQ) Analysis Guide. Available at: http://www.who.int/chp/steps/resources/GPAQ_Analysis_Guide.pdf?ua=1,2017-08-08 . Accessed 15 Mar 2017.

World Health Organization. Diet, nutrition and the prevention of chronic diseases. Report of a joint WHO/FAO expert consultation. In: WHO Technical Report Series No 916. Geneva: World Health Organization; 2003.

World Health Organization. Obesity: preventing and managing the global epidemic. In: WHO Technical Report Series 894. Geneva: World Health Organization; 2000.

Consultation WE. Appropriate body-mass index for Asian populations and its implications for policy and intervention strategies. Lancet. 2004;363:157–63.

Ahmad OB, Boschi Pinto C, D. Lopez A, Jl Murray C, Lozano R, Inoue M. Age Standardization of Rates: A New WHO Standard, vol. No. 31; 2001.

He W, Muenchrath MN, Kowal P, U.S. Census Bureau. Shades of Gray: A Cross-Country Study of Health and Well-Being of the Older Populations in SAGE Countries, 2007–2010. Washington, DC: U.S. Government Printing Office; 2012.

Roth GA, Huffman MD, Moran AE, Feigin V, Mensah GA, Naghavi M, et al. Global and regional patterns in cardiovascular mortality from 1990 to 2013. Circulation. 2015;132:1667–78.

GBD 2016 Causes of Death Collaborators. Global, Regional, and National Age-Sex Specific Mortality for 264 Causes of Death, 1980–2016: A Systematic Analysis for the Global Burden of Disease Study 2016. Lancet. 2017;390(10100):1151–210.

Omran AR. The epidemiologic transition: a theory of the epidemiology of population change. Milbank Mem Fund Q. 1971;49:509–38.

Yusuf S, Hawken S, Ounpuu S, Dans T, Avezum A, Lanas F, et al. Effect of potentially modifiable risk factors associated with myocardial infarction in 52 countries (the INTERHEART study): case-control study. Lancet. 2004;364:937–52.

O'Donnell MJ, Xavier D, Liu L, Zhang H, Chin SL, Rao-Melacini P, Rangarajan S, et al. Risk factors for ischaemic and intracerebral haemorrhagic stroke in 22 countries (the INTERSTROKE study.): a case-control study. Lancet. 2010;376:112–23.

Danaei G, Finucane MM, Lin JK, Singh GM, Paciorek CJ, Cowan MJ, et al. National, regional, and global trends in systolic blood pressure since 1980: systematic analysis of health examination surveys and epidemiological studies with 786 country-years and 5.4 million participants. Lancet. 2011;377:568–77.

Ettehad D, Emdin CA, Kiran A, Anderson SG, Callender T, Emberson J, et al. Blood pressure lowering for prevention of cardiovascular disease and death: a systematic review and meta-analysis. Lancet. 2016;387(10022):957–67.

Lloyd-Sherlock P, Beard J, Minicuci N, Ebrahim S, Chatterji S. Hypertension among older adults in low- and middle-income countries: prevalence, awareness and control. Int J Epidemiol. 2014;43:116–28.

Roberts B, Stickley A, Balabanova D, Haerpfer C, McKee M. The persistence of irregular treatment of hypertension in the former Soviet Union. J Epidemiol Community Health. 2012;66:1079–82.

Su C, Huang H, Wang HJ, Wang ZH, Zhang JG, Du WW, et al. Study on status and trend of physical activity among Chinese adults aged 18-49 years old in 9 provinces from 1997 to 2009. Chin J Health Educ. 2013;186:5926–32.

Li Y, Wang DD, Ley SH, Howard AG, He Y, Lu Y, et al. Potential impact of time trend of life-style factors on cardiovascular disease burden in China. J Am Coll Cardiol. 2016;68:818–33.

NCD Risk Factor Collaboration (NCD-RisC). Trends in adult body-mass index in 200 countries from 1975 to 2014: a pooled analysis of 1698 population-based measurement studies with 19·2 million participants. Lancet. 2016;387:1377–96.

Ng M, Fleming T, Robinson M, Thomson B, Graetz N, Margono C, et al. Global, regional, and national prevalence of overweight and obesity in children and adults during 1980-2013: a systematic analysis for the Global Burden of Disease Study 2013. Lancet. 2014;384:766–81.

Ma J, Cai XH, Wang HJ, Dong B, Song Y, Hu PJ, et al. The trend analysis of overweight and obesity in Chinese students during 1985–2010. Chin J Prev Med. 2012;46:776–80.

Hall JN, Moore S, Harper SB, Lynch JW. Global variability in fruit and vegetable consumption. Am J Prev Med. 2009;36:402–9.

Wu F, Guo Y, Chatterji S, Zheng Y, Naidoo N, Jiang Y, et al. Common risk factors for chronic non-communicable diseases among older adults in China, Ghana, Mexico, India, Russia and South Africa: the study on global AGEing and adult health (SAGE) wave 1. BMC Public Health. 2015; https://doi.org/10.1186/s12889-015-1407-0 .

Patel V, Chatterji S, Chisholm D, Ebrahim S, Gopalakrishna G, Mathers C, et al. Chronic diseases and injuries in India. Lancet. 2011;377:413–28.

Institute of Health Metrics and Evaluation. GBD Compare 2010. Available at: http://vizhub.healthdata.org/gbd-compare/ . Accessed 30 Apr 2017.

Giovino GA, Mirza SA, Samet JM, Gupta PC, Jarvis MJ, Bhala N, et al. Tobacco use in 3 billion individuals from 16 countries: an analysis of nationally representative cross-sectional household surveys. Lancet. 2012;380:668–79.

Bhan N, Srivastava S, Agrawal S, Subramanyam M, Millett C, Selvaraj S, et al. Are socioeconomic disparities in tobacco consumption increasing in India? A repeated cross-sectional multilevel analysis. BMJ Open. 2012. https://doi.org/10.1136/bmjopen-2012-001348 .

Fernández-Solà J. Cardiovascular risks and benefits of moderate and heavy alcohol consumption. Nat Rev Cardiol. 2015;12:576–87.

Roerecke M, Rehm J. Alcohol consumption, drinking patterns, and ischemic heart disease: a narrative review of meta-analyses and a systematic review and meta-analysis of the impact of heavy drinking occasions on risk for moderate drinkers. BMC Med. 2014;12:182.

Britton A, McKee M. The relation between alcohol and cardiovascular disease in Eastern Europe: explaining the paradox. J Epidemiol Community Health. 2000;54:328–32.

Article PubMed PubMed Central CAS Google Scholar

Leong DP, Smyth A, Teo KK, McKee M, Rangarajan S, Pais P, et al. Patterns of alcohol consumption and myocardial infarction risk: observations from 52 countries in the INTERHEART case control study. Circulation. 2014;130:390–8.

Bell S, Daskalopoulou M, Rapsomaniki E, George J, Britton A, Bobak M, et al. Association between clinically recorded alcohol consumption and initial presentation of 12 cardiovascular diseases: population based cohort study using linked health records. BMJ. 2017. https://doi.org/10.1136/bmj.j909 .

Download references

Acknowledgements

The authors would like to thank the respondents and interviewers from all six SAGE countries for their contributions and hard work.

This work was supported by WHO, the US National Institutes on Aging through Interagency Agreements [OGHA 04034785; YA1323-08-CN-0020; Y1-AG-1005-01] and through a research grant (R01-AG034479), and Three-year Action Plan on Public Health, Phase IV, Shanghai, China[15GWZK0801;GWIV-22].

Availability of data and materials

The datasets supporting the conclusions of this article are available upon request in the website of WHO ( http://apps.who.int/healthinfo/systems/surveydata/index.php/catalog/sage ).

Author information

Ye Ruan and Yanfei Guo contributed equally to this work.

Authors and Affiliations

Shanghai Municipal Center for Disease Control and Prevention (Shanghai CDC), Shanghai, China

Ye Ruan, Yanfei Guo, Yang Zheng, Zhezhou Huang, Shuangyuan Sun, Yan Shi & Fan Wu

World Health Organization, Geneva, Switzerland

Research Institute for Health Sciences, Chiang Mai University, Chiang Mai, Thailand

You can also search for this author in PubMed Google Scholar

Contributions

FW, PK, YFG and YZ designed, implemented the conduct of this study. YR and YFG conceived of the analysis, and drafted the manuscript. YR, YFG, YS, ZZH, YZ and SYS contributed to the statistical analyses. ZZH and SYS contributed to the editing of initial draft. All authors read and approved the final manuscript.

Corresponding authors

Correspondence to Yan Shi or Fan Wu .

Ethics declarations

Ethics approval and consent to participate.

WHO’s Ethical Review Committee approved SAGE (RPC146), and each country obtained local ethical approval to conduct the study. Written informed consent was obtained from each respondent.

Competing interests

The authors declare that they have no competing interests.

Publisher’s Note

Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations.

Rights and permissions

Open Access This article is distributed under the terms of the Creative Commons Attribution 4.0 International License ( http://creativecommons.org/licenses/by/4.0/ ), which permits unrestricted use, distribution, and reproduction in any medium, provided you give appropriate credit to the original author(s) and the source, provide a link to the Creative Commons license, and indicate if changes were made. The Creative Commons Public Domain Dedication waiver ( http://creativecommons.org/publicdomain/zero/1.0/ ) applies to the data made available in this article, unless otherwise stated.

Reprints and permissions

About this article

Cite this article.

Ruan, Y., Guo, Y., Zheng, Y. et al. Cardiovascular disease (CVD) and associated risk factors among older adults in six low-and middle-income countries: results from SAGE Wave 1. BMC Public Health 18 , 778 (2018). https://doi.org/10.1186/s12889-018-5653-9

Download citation

Received : 04 December 2017

Accepted : 01 June 2018

Published : 20 June 2018

DOI : https://doi.org/10.1186/s12889-018-5653-9

Share this article

Anyone you share the following link with will be able to read this content:

Sorry, a shareable link is not currently available for this article.

Provided by the Springer Nature SharedIt content-sharing initiative

Cardiovascular diseases
Risk Factors
Low and middle income countries

BMC Public Health

ISSN: 1471-2458

General enquiries: [email protected]

Open access
Published: 10 October 2023

Early prediction of heart disease with data analysis using supervised learning with stochastic gradient boosting

Anil Pandurang Jawalkar 1 ,
Pandla Swetcha 1 ,
Nuka Manasvi 1 ,
Pakki Sreekala 1 ,
Samudrala Aishwarya 1 ,
Potru Kanaka Durga Bhavani 1 &
Pendem Anjani 1

Journal of Engineering and Applied Science volume 70 , Article number: 122 ( 2023 ) Cite this article

3469 Accesses

5 Citations

Metrics details

Heart diseases are consistently ranked among the top causes of mortality on a global scale. Early detection and accurate heart disease prediction can help effectively manage and prevent the disease. However, the traditional methods have failed to improve heart disease classification performance. So, this article proposes a machine learning approach for heart disease prediction (HDP) using a decision tree-based random forest (DTRF) classifier with loss optimization. Initially, preprocessing of the dataset with patient records with known labels is performed for the presence or absence of heart disease records. Then, train a DTRF classifier on the dataset using stochastic gradient boosting (SGB) loss optimization technique and evaluate the classifier’s performance using a separate test dataset. The results demonstrate that the proposed HDP-DTRF approach resulted in 86% of precision, 86% of recall, 85% of F1-score, and 96% of accuracy on publicly available real-world datasets, which are higher than traditional methods.

Introduction

One person dies due to cardiovascular disease every 36 s in every country. Coronary heart disease is the leading cause of mortality in the USA, accounting for one out of every four fatalities that occur each year. This disease claims the lives of about 0.66 million people annually [ 1 ]. The expenditures associated with cardiovascular disease are significant for the healthcare system in the USA. In the years 2021 and 2022, it resulted in annual costs of around $219 billion owing to the increased demand for medical treatment and medication and the loss of productivity caused by deaths. Table 1 provides the statistics of the heart disease dataset with total heart disease cases, deaths, case fatality rate, and total vaccinations. A prompt diagnosis also aids in preventing heart failure, which is another potential cause of mortality in certain cases [ 2 ]. Since many traits put a person at risk for acquiring the ailment, it is difficult to diagnose heart disease in its earlier stages while it is still in its infancy. Diabetes, hypertension, elevated cholesterol levels, an irregular pulse rhythm, and a wide variety of other diseases are some risk factors that might contribute to this [ 3 ]. These ailments are grouped and discussed under “heart disease,” an umbrella word. The symptoms of cardiac disease can differ considerably from one individual to the next and from one condition to another within the same patient [ 4 ]. The process of identifying and classifying cardiovascular diseases is a continuous one that has a chance of being fruitful when carried out by a qualified professional with appropriate knowledge and skill in the relevant sector. There are a lot of different aspects, such as age, diabetes, smoking, being overweight, and eating a diet high in junk food. There have been several variables and criteria discovered that have been shown to either cause heart disease or raise the risk of developing heart disease [ 5 ].

Most hospitals use management software to monitor the clinical and patient data they collect. It is well-known these days, and these kinds of devices generate a vast quantity of information on patients. These data are used for decision-making help in clinical settings rather seldom. These data are precious, yet a significant portion of their knowledge is left unused [ 6 ]. Because of the sheer volume of data involved in the process, the translation of clinical data that has been acquired into information that intelligent systems can use to assist healthcare practitioners in making decisions is a process fraught with difficulties [ 7 ]. Intelligent systems put this knowledge to use to enhance the quality of treatment provided to patients. As a result of this issue, research on the processing of medical photographs was carried out. Because there were not enough specialists and too many instances were misdiagnosed, an automated detection method that was both quick and effective was necessary [ 8 ].

The primary objective of the research is centered around the effective utilization of a classifier model, which aims to categorize and identify vital components within complex medical data. This categorization process is a critical step towards enabling early diagnosis of cardiovascular diseases, potentially contributing to improved patient outcomes and healthcare management [ 9 ]. However, the pursuit of disease prediction at an early stage is not without its challenges. One significant factor pertains to the inherent complexity of the predictive methods employed in the classification process [ 10 ]. The intricate nature of these methods can lead to difficulties in interpreting the underlying decision-making processes, which might impede the integration of these models into clinical practice. Furthermore, the efficiency of disease prediction models is impacted by the time they take to execute. Swift diagnosis and intervention are crucial in medical conditions, and time-intensive models might not align with the urgency required for timely medical decisions. Researchers [ 11 ] have investigated various alternative strategies to forecast cardiovascular diseases. Perfect treatment and diagnosis have the potential to save the lives of an infinite number of individuals. The novel contribution of this work is as follows:

Preprocessing of HDP dataset with normalization, exploratory data analysis (EDA), data visualization, and extraction of top correlated features.

Implementation of DTRF classifier for training preprocessed dataset, which can accurately predict the presence or absence of heart disease.

The SGB loss optimization is used to reduce the losses generated during the training process, which tunes the hyperparameters of DTRF.

The rest of the article is organized as follows: Sect. 2 gives a detailed literature survey analysis. Section 3 gives a detailed analysis of the proposed HDP-DTRF with multiple modules. Section 4 gives a detailed simulation analysis of the proposed HDP-DTRF. Section 5 concludes the article.

Literature survey

Rani et al. [ 12 ] designed a novel hybrid decision support system to diagnose cardiac ailments early. They effectively addressed the missing data challenge by employing multivariate imputations through chained equations. Additionally, their unique approach to feature selection involved a fusion of genetic algorithms (GA) and recursive feature reduction. Notably, the integration of random forest classifiers played a pivotal role in significantly enhancing the accuracy of their system. However, despite these advancements, their hybrid approach’s complexity might have posed challenges in terms of interpretability and practical implementation. Kavitha et al. [ 13 ] embraced machine learning techniques to forecast cardiac diseases. They introduced a hybrid model by incorporating random forest as the base classifier. This hybridization aimed to enhance prediction accuracy; however, their decision to capture and store user input parameters for future use was intriguing but yielded suboptimal classification performance. This unique approach could be viewed as an innovative attempt to integrate patient-specific information, yet the exact impact on overall performance warrants further investigation.

Mohan et al. [ 14 ] further advanced the field by employing a hybrid model that combined random forest with a linear model to predict cardiovascular diseases. Through this amalgamation of different classification approaches and feature combinations, they achieved commendable performance with an accuracy of 88.7%. However, it is worth noting that while hybrid models show promise, the trade-offs between complexity and interpretability could influence their practical utility in real-world clinical settings. To predict heart diseases, Shah et al. [ 15 ] adopted supervised learning techniques, including Naive Bayes, decision trees, K-nearest neighbor (KNN), and random forest algorithms. Their choice of utilizing the Cleveland database from the UCI repository as their data source added a sense of universality to their findings. However, the lack of customization in data sources might limit the applicability of their model to diverse patient populations with varying characteristics. Guoet et al. [ 16 ] contributed to the field by harnessing an improved learning machine (ILM) model in conjunction with machine learning techniques. Integrating novel feature combinations and categorization methods showcased their dedication to enhancing performance and accuracy. Nonetheless, while their approach exhibits promising results, the precise impact of specific feature combinations on prediction accuracy could have been further explored. Hager Ahmed et al. [ 17 ] presented an innovative real-time prediction system for cardiac diseases using Apache Spark and Apache Kafka. This system, characterized by its three-tier architecture—offline model building, online prediction, and stream processing pipeline—highlighted its commitment to harnessing cutting-edge technologies for practical medical applications. However, the scalability and resource requirements of such real-time systems, especially in healthcare settings with limited computational resources, could be an area of concern.

Kataria et al. [ 18 ] comprehensively analyzed and compared various machine learning algorithms for predicting heart disease. Their focus on analyzing the algorithms’ ability to predict heart disease effectively sheds light on their dedication to identifying the most suitable model. However, their study’s outcome might have been further enriched by addressing the unique challenges posed by individual attributes, such as high blood pressure and diabetes, in a more customized manner. Kannan et al. [ 19 ] meticulously evaluated machine learning algorithms to predict and diagnose cardiac sickness. By selecting 14 criteria from the UCI Cardiac Datasets, they showcased their dedication to designing a comprehensive study. Nevertheless, a deeper analysis of how these algorithms perform with specific criteria and their contributions to accurate predictions could provide more actionable insights.

Ali et al. [ 20 ] conducted a detailed analysis of supervised machine-learning algorithms for predicting cardiac disease. Their thorough evaluation of decision trees, k-nearest neighbors, and logistic regression classifiers (LRC) provided a well-rounded perspective on the strengths and limitations of each method. However, a more fine-grained analysis of how these algorithms perform under various parameter configurations and feature combinations might offer additional insights into their potential use cases. Mienye et al. [ 21 ] introduced an enhanced technique for ensemble learning, utilizing decision trees, random forests, and support vector machine classifiers. The voting system they employed to aggregate results showcased their innovative approach to combining various methods. However, the potential trade-offs between ensemble complexity and the robustness of predictions could be considered for future refinement. Dutta et al. [ 22 ] revolutionized the field by introducing convolutional neural networks (CNNs) for predicting coronary heart disease. Their approach, leveraging the power of CNNs on a large dataset of ECG signals, showcased the potential for deep learning techniques in healthcare. However, the requirement for extensive computational resources and potential challenges in model interpretability could be areas warranting further attention. Latha et al. [ 23 ] demonstrated ensemble classification approaches. Combined with a bagging technique, their utilization of decision trees, naive Bayes, and random forest exemplified their determination to achieve robust results. Nevertheless, the potential interplay between different ensemble techniques and their effectiveness under various scenarios could be explored further.

Ishaq et al. [ 24 ] introduced the concept of using the synthetic minority oversampling technique (SMOTE) in conjunction with efficient data mining methods to improve survival prediction for heart failure patients. Their emphasis on addressing class imbalance through SMOTE showcased their awareness of real-world challenges in healthcare datasets. However, the potential impact of the SMOTE method on individual patient subgroups and its implications for model fairness could be areas of future exploration. Asadi et al. [ 25 ] proposed a unique cardiac disease detection technique based on random forest swarm optimization. Their use of a large dataset for evaluation underscored their dedication to robust testing. However, the potential influence of dataset characteristics and the algorithm’s sensitivity to various parameters on prediction performance could be investigated further.

Proposed methodology

Heart disease is a significant health problem worldwide and is responsible for many deaths every year. Traditional methods for diagnosing heart disease are often time-consuming, expensive, and inaccurate. Therefore, there is a need for more accurate and efficient methods for predicting and diagnosing heart disease. The article aims to provide a detailed analysis of the proposed HDP-DTRF approach and its performance in accurately predicting the presence or absence of heart disease. The results demonstrate the effectiveness of the proposed approach, which can lead to improved diagnosis and treatment of heart disease, ultimately leading to better health outcomes for patients.

Figure 1 shows the proposed HDP-DTRF block diagram. The initial step in the proposed approach is the preprocessing of a dataset consisting of patient records with known labels indicating the presence or absence of heart disease. The dataset is then used to train a DTRF classifier with the SGB loss optimization technique. The performance of the trained classifier is evaluated using a separate publicly available real-world test dataset, and the results show that the proposed HDP-DTRF approach can accurately predict the presence or absence of heart disease. Using decision trees in the random forest classifier enables the algorithm to handle nonlinear data and make accurate predictions even with missing or noisy data. Applying the SGB loss optimization technique further enhances the algorithm’s performance by improving the convergence rate and avoiding overfitting. The proposed approach can be useful in clinical decision-making processes, enabling medical professionals to predict the likelihood of heart disease in patients and take appropriate preventive measures.

Block diagram for the proposed HDP-DTRF system

The detailed operation of the proposed HDP-DTRF system is illustrated as follows:

Step 1: Data preprocessing: Gather a dataset containing patient records, where each record includes features such as age, blood pressure, and cholesterol levels, along with labels indicating whether the patient has heart disease. Remove duplicate records, handle missing values (e.g., imputing missing data or removing instances with missing values), and eliminate irrelevant or redundant features. Encode categorical variables (like gender) into numerical values using techniques like one-hot encoding. Scale numerical features to bring them to a common scale, which can prevent features with larger ranges from dominating the model.

Step 2: Training the DTRF classifier: Initialize an empty random forest ensemble. For each tree in the ensemble, randomly sample the training data with replacement. It creates a bootstrapped dataset for training each tree, ensuring diversity in the data subsets. Construct a decision tree using the bootstrapped dataset. At each node of the tree, split the data based on the feature that provides the best separation, determined using metrics like Gini impurity or information gain. Add the constructed decision tree to the random forest ensemble. Repeat the process to create the ensemble’s desired number of decision trees.

Step 3: SGB optimization: Initialize the model by setting the initial prediction to the mean of the target labels. Calculate the negative gradient of the loss function (such as mean squared error or log loss) concerning the current model’s predictions. This gradient represents the direction in which the model’s predictions need to be adjusted to minimize the loss. Train a new decision tree using the negative gradient as the target. This new tree will help correct the errors made by the previous model iterations. Update the model’s predictions by adding the predictions of the new tree, scaled by a learning rate. This step moves the model closer to the correct predictions. Repeat the process for a predefined number of iterations. Each iteration focuses on improving the model’s predictions based on the errors made in the previous iterations.

Step 4: Performance evaluation: Use a separate real-world test dataset that was not used during training to evaluate the performance of the trained HDP-DTRF classifier.

DTRF classifier

The DTRF classifier, an ensemble learning model, centers around the decision tree as its core component. As illustrated in Fig. 2 , the DTRF block diagram depicts a framework comprising multiple trained decision trees employing the bagging technique. During the classification process, when a sample requiring classification is input, the ultimate classification outcome is determined through a majority vote from the output of an individual decision tree [ 26 ]. In classifying high-dimensional data, the DTRF model outperforms standalone decision trees by effectively addressing overfitting, displaying robust resistance to noise and outliers, and demonstrating exceptional scalability and parallel processing capabilities. Notably, the strength of DTRF stems from its inherent parameter-free nature, embodying a data-driven approach. The model requires no prior knowledge of classification from the user and is adept at training classification rules based on observed instances. This data-centric attribute enhances the model’s adaptability to various data scenarios. The DTRF model’s essence lies in utilizing K decision trees. Each of these trees contributes a single “vote” towards the category it deems most fitting, thereby participating in determining the class to which the independent variable X, under consideration, should be allocated. This approach effectively harnesses the collective wisdom of multiple trees, facilitating accurate and robust classification outcomes that capitalize on the diverse insights provided by each decision tree. The mathematical analysis of DTRF is as follows:

Block diagram of DTRF

Here, $K$ represents the number of decision trees present in the DTRF. In this context, ${\theta }_{k}$ is a collection of independent random vectors uniformly distributed amongst themselves. Here, $K$ individual decision trees are generated. Each tree provides its prediction for the category that best fits the independent variable $X$ . The predictions made by the $K$ decision trees are combined through a voting mechanism to determine the final category assignment for the independent variable $X$ . It is important to note that the given Eq. ( 1 ) indicates the ensemble nature of the DTRF model, where multiple decision trees work collectively to enhance predictive accuracy and robustness. The collection of ${\theta }_{k}$ represents the varied parameter sets for each decision tree within the ensemble.

The following procedures must be followed to produce a DTRF:

Step 1: The $K$ classification regression trees are generated by randomly selecting $K$ samples from the original training set as a self-service sample set, using the random repeated sampling method. Extracting all $K$ samples requires repeating this procedure.

Step 2: Each node in the trees will include m randomly selected characteristics from the first training set (m n). Only one of the m traits is employed in the node splitting procedure, and it is the one with the greatest classification potential. DTRF calculates how much data is included in each feature to do this.

Step 3: A tree never has to be trimmed since it grows perfectly without help.

Step 4: The generated trees are built using DTRFs, and the freshly received data is categorized using DTRFs. The number of votes from the tree classifiers determines the classification outcomes.

There are a lot of important markers of generalization performance that are inherent to DTRFs. Similarity and correlation between different decision trees, mistakes in generalization, and the system’s ability to generalize are all features t . A system’s decision-making efficacy is determined by how well it can generalize its results to fresh information that follows the same distribution as the training set [ 27 ]. The system’s performance and generalizability benefit from reducing the severity of generalization mistakes. Here is a case of the overgeneralization fallacy in action:

Here, $P{E}^{*}$ denotes the generalization error, the subscripts $X$ and $Y$ point to the space where the probability is defined, and $Mr (X, Y)$ is the margin function. The following is a definition of the margin function:

If it stands for the input sample, $Y$ indicates the correct classification, and $J$ indicates the incorrect one. Specifically, $h(g)$ is a representation of a sequence model for classification, $I(g)$ indicates an indicator function, and ${avg}_{k}(g)$ means averaging. The margin function determines how many more votes the correct classification for sample X receives than all possible incorrect classifications. As the value of the margin function grows, so does the classifier’s confidence in its accuracy. The term “convergence formulation of generalization error” as follows [ 28 ]:

As the number of decision trees grows, the generalization error will tend toward a maximum, as predicted by the preceding calculation, and the model will not over-fit. The classification power of each tree and the correlation between trees is used to estimate the maximum allowed generalization error. The DTRF model aims to produce a DTRF with a small correlation coefficient and strong classification power. Classification intensity ( $S$ ) is the sample-space-wide mathematical expectation of the variable $mr(X, Y)$ .

Here, $\theta$ and $\theta {\prime}$ are independent and identically distributed vectors of estimated data ${E}_{X, Y}$ , correlation coefficients of $mr(\theta , X, Y)$ and $mr(\theta ,{\prime} X, Y)$ :

Among them, $sd(\theta )$ can be expressed as follows:

Equation ( 7 ) is a metric that is used to quantify the degree to which the trees $h(X,\theta )$ and $h(X,\theta {\prime})$ on the dataset consisting of X , Y are correlated with one another. The correlation coefficient increases in magnitude in direct proportion $\overline{\rho }$ to the size of the chi-square. The upper limit of generalization error is obtained using the following formula, which is based on the Chebyshev inequality:

The generalization error limit of a DTRF is inversely proportional to the strength of the correlation P between individual decision trees and positively correlated with the classification intensity S of a single tree. That is to say, the stricter the category $S$ , the lower the degree of linkage $P$ . If the DTRF is to improve its classification accuracy, the threshold for generalization error must be lowered.

SGB loss optimization

The SGB optimization approach has recently received increased use in various deep-learning applications. These applications call for a higher degree of expertise in learning than what can be provided by more conventional means. During the whole training process, the learning rate that SGB uses does not, at any time, experience any fluctuations. The SGB uses one learning rate, which is alpha. The SGB algorithm maintains a per-parameter learning rate to increase performance in scenarios with sparse gradients (for example, computer vision challenges). It maintains per-parameter learning rates that are updated based on the average of recent magnitudes of the gradients for the weight, and it does so based on averaging recent gradient magnitudes (for example, how rapidly it is changing). In addition, it does this based on averaging recent gradient magnitudes for the weight. It illustrates that the strategy is effective for online and non-stationary applications (for example, noisy). The chain rule applied calculus to compute the partial derivatives. To calculate the loss gradient about the weights and biases, it will allow us to determine how the loss varies as a function of the weights and biases. Let us assume that we have a training dataset with N samples, denoted as { ${x}_{i}, {y}_{i}$ } for i = 1, 2, …, N , where ${x}_{i}$ is the input, and ${y}_{i}$ is the true label or target value. It uses a decision tree with parameters θ to predict the output ${\widehat{\mathrm{y}}}_{i}$ for input ${x}_{i}$ . The output can be any function of the parameters and the input, represented as ${\widehat{\mathrm{y}}}_{i} = f({x}_{i}, \theta ).$ The goal is to minimize the difference between the predicted output ${\widehat{\mathrm{y}}}_{i}$ and the true label ${y}_{i}$ . It is typically done by defining a loss function $L({\widehat{\mathrm{y}}}_{i}, {y}_{i})$ that quantifies the difference between the predicted and true values. The total loss over the entire dataset is then defined as the sum of the individual losses over all samples:

The optimization algorithm focused on estimating the values of the parameters $\theta$ that minimize this total loss. It is typically done using gradient descent, which updates the parameters $\theta$ in the opposite direction of the gradient of the total loss concerning the parameters:

Here, $\alpha$ is the learning rate, which controls the size of the parameter update, and ${\nabla }_{\theta }{L}_{total}$ is the gradient of the total loss concerning the parameters θ. The SGB can sometimes oscillate and take a long time to converge due to the noisy gradients. Momentum is a technique that helps SGB converge faster by adding a fraction of the previous update to the current update:

Here, ${v}_{t}$ is the momentum term at iteration $t, \beta$ is the momentum coefficient, typically set to 0.9 or 0.99, and the other terms are as previously defined.

Results and discussion

This section gives a detailed performance analysis of the proposed HDP-DTRF. The performance of the proposed method is measured using multiple performance metrics. All these metrics are measured for proposed methods as well as existing methods. Then, all the methods use the same publicly available real-world dataset for performance estimations.

The Cleveland Heart Disease dataset contains data on 303 patients who were evaluated for heart disease. The dataset is downloaded from open-access websites like the UCI-ML repository. Each patient is represented by 14 attributes, which include demographic and clinical information such as age, sex, chest pain type, resting blood pressure, serum cholesterol level, and exercise test results. The dataset has 303 records, each corresponding to a unique patient. The data in each record includes values for all 14 attributes, and the diagnosis of heart disease (present or absent) is also included in the dataset. Table 2 provides a detailed description of the dataset. Researchers and data scientists can use this dataset to develop predictive models for heart disease diagnosis or explore relationships between the different variables in the dataset. With 303 records, this dataset is relatively small compared to other medical datasets. However, it is still widely used in heart disease research due to its rich attributes and long history of use in research studies.

EDA is essential in understanding and analyzing any dataset, including the Cleveland Heart Disease dataset. EDA involves examining the dataset’s basic properties, identifying missing values, checking data distributions, and exploring relationships between variables. Figure 3 shows the EDA of the dataset. Figure 3 (a) shows the count for each target class. Here, the no heart disease class contains 138 records, and the heart disease presented class contains 165 records. Figure 3 (b) shows the male and female-based record percentages in the dataset. Here, the dataset contains 68.32% male and 31.68% female records. Figure 3 (c) shows the percentage of records for chest pain experienced by the patient in the dataset. Here, the dataset contains 47.19% of records in typical angina, 16.50% in atypical angina, 28.71% in non-anginal pain, and 7.59% in the asymptomatic class. Figure 3 (d) shows the percentage of records for fasting blood sugar in the dataset. Here, the dataset contains 85.15% of records in the fasting blood sugar (> 120 mg/dl) class and 14.85% of records in the fasting blood sugar (< 120 mg/dl) class. Figure 4 shows the heart disease frequency by age for both no disease and disease classes. The output contains histogram levels that show the frequency of heart disease by age. Here, the counts of patients with and without heart disease are shown in red and green colors. The overlap between the bars shows how the frequency of heart disease varies with age, with a peak in the frequency of heart disease occurring around the age of 29–77 years.

EDA of the dataset. a Count for each target class. b Male–female distribution. c Chest pain experienced by patient distribution. d Fasting blood sugar distribution

Heart disease frequency by age

Figure 5 shows the frequencies for different columns of the dataset, which contains the frequencies of chest pain type, fasting blood sugar, rest ECG, exercise-induced angina, st_slope, and number of major vessel columns. Exploring the frequencies of different variables in a dataset is crucial in understanding the data and gaining insights about the underlying phenomena. By analyzing the frequency of values in each variable, we can better understand the data distribution and identify potential patterns, relationships, and outliers that are important for further analysis. For example, understanding the frequency of different chest pain types in a heart disease dataset reveals whether certain types of chest pain are more strongly associated with the disease than others. Similarly, analyzing the frequency of different fasting blood sugar levels helps to identify potential risk factors for heart disease. Overall, exploring the frequencies of variables is an important step in the EDA process, as it provides a starting point for identifying potential relationships and patterns in the data.

Frequencies for different columns of the dataset

Performance evaluation

Table 3 shows the class-specific performance evaluation of HDP-DTRF. Here, the performance was measured for class-0 (no heart disease) and class-1 (heart disease presented) classes. Further, macro average and weighted average performances were also measured. Macro average treats all classes equally, regardless of their size. It calculates the average performance metrics across all classes, giving each class an equal weight. It means that the performance of smaller classes will have the same impact on the metric as larger classes.Then, the weighted average considers the size of each class. It calculates the average performance metric across all classes but gives each class a weight proportional to its size. It means that the performance of larger classes will have a greater impact on the metric than smaller classes.

Table 4 shows the class-0 performance comparison of various methods. Here, the proposed HDP-DTRF improved precision by 5.75%, recall by 1.37%, F1-score by 6%, and accuracy by 2.45% compared to KNN [ 15 ]. Then, the proposed HDP-DTRF improved precision by 3.45%, recall by 0.63%, F1-score by 3.61%, and accuracy by 1.45% compared to ILM [ 16 ]. Then, the proposed HDP-DTRF improved precision by 2.30%, recall by 1.27%, F1-score by 3.61%, and accuracy by 1.03% compared to LRC [ 20 ]. Table 5 shows the class-1 performance comparison of various methods. Here, KNN [ 15 ] shows a 2.35% lower precision, a 4.40% lower recall, a 3.53% lower F1-score, and a 1.03% lower accuracy than the proposed HDP-DTRF method. Then, ILM shows a 2.35% lower precision, a 5.49% lower recall, a 1.14% lower F1-score, and a 1.03% lower accuracy than the proposed HDP-DTRF method. Then, LRC [ 20 ] shows a 4.71% lower precision, an 11.11% lower recall, a 2.27% lower F1-score, and a 1.03% lower accuracy than the proposed HDP-DTRF method.

Table 6 shows the macro average performance comparison of various methods. For KNN [ 15 ], the percentage improvements are 7.5% for precision, 13.3% for recall, 10.4% for F1-score, and 6.7% for accuracy. For ILM [ 16 ], the percentage improvements are achieved as 2.4% for precision, 6.1% for recall, 6.0% for F1-score, and 3.2% for accuracy. For LRC [ 20 ], the percentage improvements are achieved as 3.4% for precision, 10.0% for recall, 6.0% for F1-score, and 4.3% for accuracy archived by the proposed method. Table 7 shows the weighted average performance comparison of various methods. For KNN [ 15 ], the percentage improvements are 6.5% for precision, 3.3% for recall, 1.4% for F1-score, and 6.7% for accuracy. For ILM [ 16 ], the percentage improvements are achieved as 2.4% for precision, 5.1% for recall, 6.0% for F1-score, and 3.2% for accuracy. For LRC [ 20 ], the percentage improvements are achieved as 1.4% for precision, 1.0% for recall, 6.0% for F1-score, and 4.3% for accuracy archived by the proposed method.

The ROC curve of the proposed HDP-DTRF is seen in Fig. 6 . The true positive rate (TPR) is shown against the false-positive rate (FPR) on the ROC curve, which considers various threshold values. In the context of the HDP-DTRF technique, the ROC curve illustrates the degree to which the model can differentiate between positive and negative heart disease instances. The model’s performance is greater when it has a higher TPR and a lower FPR. The ROC curve that represents the HDP-DTRF approach that has been suggested is used to find the best classification threshold, which strikes a balance between sensitivity and specificity in the diagnostic process. If there is a point on the ROC curve that is closer to the top left corner, this implies that the model is doing better.

ROC curve of proposed HDP-DTRF

Conclusions

This article proposes a machine-learning approach for heart disease prediction. The approach uses a DTRF classifier with loss optimization and involves preprocessing a dataset of patient records to determine the presence or absence of heart disease. The DTRF classifier is then trained on the SGB loss optimization dataset and evaluated using a separate test dataset. The proposed HDP-DTRF improved class-specific performances and a macro with weighted average performance measures. Overall, the proposed HDP-DTRF improved precision by 2.30%, recall by 1.27%, F1-score by 3.61%, and accuracy by 1.03% compared to traditional methodologies. Further, this work can be extended with deep learning-based classification with machine learning feature analysis .

Availability of data and materials

Data sharing is not applicable to this article as no datasets were generated or analyzed during the current study.

Abbreviations

Heart disease prediction

Decision tree-based random forest

Stochastic gradient boosting

False positive

False negative

True negative

True positive

Bhatt CM et al (2023) Effective heart disease prediction using machine learning techniques. Algorithms 16(2):88

Article Google Scholar

Dileep P et al (2023) An automatic heart disease prediction using cluster-based bi-directional LSTM (C-BiLSTM) algorithm. Neural Comput Appl 35(10):7253–7266

Jain A et al (2023) Optimized levy flight model for heart disease prediction using CNN framework in big data application. Exp Syst Appl 223:119859

Nandy S et al (2023) An intelligent heart disease prediction system based on swarm-artificial neural network. Neural Comput Appl 35(20):14723–14737

Hassan D et al (2023) Heart disease prediction based on pre-trained deep neural networks combined with principal component analysis. Biomed Signal Proc Contr 79:104019

Ozcan M et al (2023) A classification and regression tree algorithm for heart disease modeling and prediction. Healthc Anal 3:100130

Saranya G et al (2023) A novel feature selection approach with integrated feature sensitivity and feature correlation for improved heart disease prediction. J Ambient Intell Humaniz Comput 14(9):12005–12019

Sudha VK et al (2023) Hybrid CNN and LSTM network for heart disease prediction. SN Comp Sc 4(2):172

Chaurasia V, et al (2023) Novel method of characterization of heart disease prediction using sequential feature selection-based ensemble technique. Biomed Mat Dev 2023;1–10. https://doi.org/10.1007/s44174-022-00060-x

Ogundepo EA et al (2023) Performance analysis of supervised classification models on heart disease prediction. Innov Syst Software Eng 19(1):129–144

de Vries S et al (2023) Development and validation of risk prediction models for coronary heart disease and heart failure after treatment for Hodgkin lymphoma. J Clin Oncol 41(1):86–95

Vijaya Kishore V, Kalpana V (2020) Effect of Noise on Segmentation Evaluation Parameters. In: Pant, M., Kumar Sharma, T., Arya, R., Sahana, B., Zolfagharinia, H. (eds) Soft Computing: Theories and Applications. Advances in Intelligent Systems and Computing, vol 1154. Springer, Singapore. https://doi.org/10.1007/978-981-15-4032-5_41 .

Kalpana V, Vijaya Kishore V, Praveena K (2020) A Common Framework for the Extraction of ILD Patterns from CT Image. In: Hitendra Sarma, T., Sankar, V., Shaik, R. (eds) Emerging Trends in Electrical, Communications, and Information Technologies. Lecture Notes in Electrical Engineering, vol 569. Springer, Singapore. https://doi.org/10.1007/978-981-13-8942-9_42

Annamalai M, Muthiah P (2022) An Early Prediction of Tumor in Heart by Cardiac Masses Classification in Echocardiogram Images Using Robust Back Propagation Neural Network Classifier. Brazilian Archives of Biology and Technology. 65. https://doi.org/10.1590/1678-4324-2022210316

Shah D et al (2020) Heart disease prediction using machine learning techniques. SN Comput Sci 1:345

Guo C et al (2020) Recursion enhanced random forest with an improved linear model (RERF-ILM) for heart disease detection on the internet of medical things platform. IEEE Access 8:59247–59256

Ahmed H et al (2020) Heart disease identification from patients’ social posts, machine learning solution on Spark. Future Gen Comp Syst 111:714–722

Katarya R et al (2021) Machine learning techniques for heart disease prediction: a comparative study and analysis. Health Technol 11:87–97

Kannan R et al (2019) Machine learning algorithms with ROC curve for predicting and diagnosing the heart disease. Springer, Soft Computing and Medical Bioinformatics

Book Google Scholar

Ali MM et al (2021) Heart disease prediction using supervised machine learning algorithms: Performance analysis and comparison. Comput Biol Med 136:104672

Mienye ID et al (2020) An improved ensemble learning approach for the prediction of heart disease risk. Inform Med Unlocked 20:100402

Dutta A et al (2020) An efficient convolutional neural network for coronary heart disease prediction. Expert Syst Appl 159:113408

Latha CBC et al (2019) Improving the accuracy of heart disease risk prediction based on ensemble classification techniques. Inform Med Unlocked 16:100203

Ishaq A et al (2021) Improving the prediction of heart failure patients’ survival using SMOTE and effective data mining techniques. IEEE Access 9:39707–39716

Asadi S et al (2021) Random forest swarm optimization-based for heart diseases diagnosis. J Biomed Inform 115:103690

Asif D et al (2023) Enhancing heart disease prediction through ensemble learning techniques with hyperparameter optimization. Algorithms 16(6):308

David VAR S, Govinda E, Ganapriya K, Dhanapal R, Manikandan A (2023) "An Automatic Brain Tumors Detection and Classification Using Deep Convolutional Neural Network with VGG-19," 2023 2nd International Conference on Advancements in Electrical, Electronics, Communication, Computing and Automation (ICAECA), Coimbatore, India, 2023, pp. 1-5. https://doi.org/10.1109/ICAECA56562.2023.10200949

Radwan M et al (2023) MLHeartDisPrediction: heart disease prediction using machine learning. J Comp Commun 2(1):50-65

Download references

Acknowledgements

Not applicable.

No funding was received by any government or private concern.

Author information

Authors and affiliations.

Department of Information Technology, Malla Reddy Engineering College for Women (UGC-Autonomous), Maisammaguda, Hyderabad, India

Anil Pandurang Jawalkar, Pandla Swetcha, Nuka Manasvi, Pakki Sreekala, Samudrala Aishwarya, Potru Kanaka Durga Bhavani & Pendem Anjani

You can also search for this author in PubMed Google Scholar

Contributions

A.P.J, P.S., and N.M. contributed to the technical content of the paper, and P.S. and S.A. contributed to the conceptual content and architectural design. P.K., D.B., and P.A. contributed to the guidance and counseling on the writing of the paper.

Corresponding author

Correspondence to Anil Pandurang Jawalkar .

Ethics declarations

Competing interests.

The authors declare that they have no competing interests.

Additional information

Publisher’s note.

Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations.

Rights and permissions

Open Access This article is licensed under a Creative Commons Attribution 4.0 International License, which permits use, sharing, adaptation, distribution and reproduction in any medium or format, as long as you give appropriate credit to the original author(s) and the source, provide a link to the Creative Commons licence, and indicate if changes were made. The images or other third party material in this article are included in the article's Creative Commons licence, unless indicated otherwise in a credit line to the material. If material is not included in the article's Creative Commons licence and your intended use is not permitted by statutory regulation or exceeds the permitted use, you will need to obtain permission directly from the copyright holder. To view a copy of this licence, visit http://creativecommons.org/licenses/by/4.0/ . The Creative Commons Public Domain Dedication waiver ( http://creativecommons.org/publicdomain/zero/1.0/ ) applies to the data made available in this article, unless otherwise stated in a credit line to the data.

Reprints and permissions

About this article

Cite this article.

Jawalkar, A.P., Swetcha, P., Manasvi, N. et al. Early prediction of heart disease with data analysis using supervised learning with stochastic gradient boosting. J. Eng. Appl. Sci. 70 , 122 (2023). https://doi.org/10.1186/s44147-023-00280-y

Download citation

Received : 31 May 2023

Accepted : 05 September 2023

Published : 10 October 2023

DOI : https://doi.org/10.1186/s44147-023-00280-y

Share this article

Anyone you share the following link with will be able to read this content:

Sorry, a shareable link is not currently available for this article.

Provided by the Springer Nature SharedIt content-sharing initiative

Heart disease
Machine learning
Decision tree
Random forest
Loss optimization

Share this —

Watch Full Episodes
Read With Jenna
Inspirational
Relationships
TODAY Table
Newsletters
Start TODAY
Shop TODAY Awards
Citi Concert Series
Listen All Day

Follow today

More Brands

On The Show
TODAY Plaza

Signs of heart disease may be more subtle in women than men

Symptoms of heart disease — the country’s No. 1 killer — may be more subtle and varied in women than in men , according to a review published Thursday in the American Heart Association journal Circulation .

Understanding the differences in symptoms is particularly important for women. Corrine Jurgens, an author of the review and an associate professor at the Connell School of Nursing at Boston College, said that women tend to be diagnosed with heart disease later in life than men, when they may have other underlying conditions that could make identifying subtle symptoms of heart disease much more difficult.

What’s more, a 2020 report , also published in Circulation, found a 10-year decline in awareness among women that heart disease is indeed their biggest health threat.

“Many women are concerned about their breast cancer risk, and they perceive that as their greatest health threat,” said Dr. Deirdre Mattina, a cardiologist at the Cleveland Clinic. But “we know that one in three women are going to die of heart disease” every year.

For both women and men, signs of heart problems rarely occur in isolation.

“Symptoms often occur in clusters,” Jurgens said. “Very rarely does someone come in with just one symptom.”

And though sudden cardiac events — heart attack or stroke, for example — certainly appear without warning, many symptoms worsen over time.

Mattina said that patients with heart failure, for example, may report no longer being able to walk as far as they used to, or a gradual decline in the ability to take in full breaths.

“We’re looking for a pattern,” Mattina said.

Here are the most common ways for six different cardiovascular conditions that present in patients.

Heart attack

A heart attack occurs when blood flow to the heart muscle is cut off or drastically reduced.

Chest pain is the classic symptom, but other symptoms associated with heart attacks can be much more subtle, such as a pressure or tightness within the chest that can sometimes radiate to the jaw, arms and back.

Men are about twice as likely as women to have a heart attack.

But women often have more symptoms that accompany a heart attack than men, including nausea, lightheadedness, extreme fatigue and cold sweats.

Younger women, generally considered as younger than 55, tend to experience at least three symptoms during a heart attack. Those can include pain in the jaw, neck, arms or shoulders, chest palpitations or feelings of heartburn.

A stroke occurs when blood flow to the brain is cut off or drastically reduced.

The signs of a stroke are facial drooping, arm weakness, difficulty speaking, confusion and dizziness. Immediate medical care is necessary when a person is having a stroke.

Women tend to experience additional symptoms, such as headache, and a more severe altered mental state, according to the review.

It’s critical for patients to follow up with their physicians following a stroke, as it can affect cognitive function. That may make it more difficult for patients to identify any new symptoms.

Heart failure

Shortness of breath is the most common symptom associated with heart failure, which happens when the heart isn’t pumping blood as well as it should be. It can often occur after someone has a heart attack.

Jurgens said that heart failure symptoms may build slowly over the course of up to three weeks before people may realize they are having a problem that requires emergency medical care.

Symptoms can include upset stomach, vomiting, loss of appetite, fatigue, mood changes and trouble with memory.

Women with heart failure have a wider variety of symptom, such as sweating, unusual swelling, heart palpitations and feelings of heartburn. Those symptoms, the review found, are often accompanied by depression and anxiety.

Depression tends to be more prevalent in people with heart failure and other cardiovascular conditions. According to the review, 10% of people with heart disease experience depression, compared with 5% of those without heart problems.

That may make it difficult for patients to figure out whether symptoms — fatigue, for example — are due to depression or heart disease or both.

Heart valve disease

Heart valve disease occurs when one or more of valves in the heart doesn’t work properly. Like heart failure, shortness of breath is often reported.

It can lead to a complication called aortic stenosis, which occurs when the valve that allows blood to flow from the heart to the rest of the body is narrowed, restricting that blood flow. While men are more likely to experience chest pain with valve disease, women tend to report more trouble catching their breath and exercising.

Abnormal heart rhythm

An arrhythmia, or irregular heartbeat, occurs when the heart’s electrical signals misfire, making the heart beat too quickly or too slowly.

The problem is often felt as fluttering in the chest, especially in women.

Men often don’t experience symptoms of an irregular heartbeat at all.

But sometimes fatigue, chest pain, shortness of breath and dizziness can also accompany an arrhythmia. Black Americans are most likely to report those symptoms.

Peripheral vascular disease

The legs are not to be ignored when it comes to heart disease risks and their accompanying symptoms.

Peripheral vascular disease can lead to amputation and can increase the risk for heart attack and stroke.

There are two types: peripheral artery disease and peripheral vein disease.

Peripheral vein disease affects blood flow from the legs back to the heart, and can lead to blood clots and deep vein thrombosis.

Peripheral artery disease occurs when cholesterol builds up in the arteries that carry blood to the extremities, usually the legs.

For peripheral artery disease, “one of the primary symptoms is difficulty walking,” said Dr. Amy Pollak, a cardiologist at the Mayo Clinic. Though leg and foot pain can occur, she said that many patients report leg fatigue, and sometimes heaviness or discomfort in their legs.

It’s “a symptom that many patients chalk up to something else,” Pollak said. “They think it’s arthritis or neuropathy or aging.”

Indeed, women may also have accompanying conditions, such as osteoarthritis, that could mirror or mask symptoms of peripheral artery disease.

Figuring out the cause of such leg pain or discomfort is key, Pollak said, as it “may be a really important clue to that greater arterial tree that runs through our body, connecting our heart, brain and legs.”

This story originally appeared on NBCNews.com .

These foods can reduce pain from period cramps

Women's health.

What is cycle syncing? Everything to know, from foods to workouts to fertility

Jennifer Lopez and Ben Affleck split: What’s your divorce risk by years of marriage?

I was battling cancer while starring on Broadway and no one knew

Woman, 106, still lives in her own apartment, eats these healthy foods every day

Common endometrial cancer test is less effective for Black women, research shows

1 exercise alerts woman to 'scary' weakness after extreme reaction to cholesterol drug

Woman shocked when eye freckle turns into melanoma

During labor, her heart stopped for 30 seconds. She had a rare reaction to her amniotic fluid

College student, 20, experienced numbness in hands, feet. It helped doctors find ovarian cancer

An official website of the United States government

The .gov means it’s official. Federal government websites often end in .gov or .mil. Before sharing sensitive information, make sure you’re on a federal government site.

The site is secure. The https:// ensures that you are connecting to the official website and that any information you provide is encrypted and transmitted securely.

Publications
Account settings

Preview improvements coming to the PMC website in October 2024. Learn More or Try it out now .

Advanced Search
Journal List
Med J Armed Forces India
v.77(3); 2021 Jul

Machine learning–based heart disease prediction system for Indian population: An exploratory study done in South India

a Research Scholar (Computer Science & Engineering), Dayananda Sagar University, Bengaluru, India

Bondu Venkateswarlu

b Associate Professor (Computer Science & Engineering), Dayananda Sagar University, Bengaluru, India

Baljeet Maini

c Professor Pediatrics, Teerthanker Mahaveer Medical College & Research Centre, Moradabad, India

Dheeraj Marwaha

d Senior Software Engineer, Microsoft India, Hyderabad, India

In India, huge mortality occurs due to cardiovascular diseases (CVDs) as these diseases are not diagnosed in early stages. Machine learning (ML) algorithms can be used to build efficient and economical prediction system for early diagnosis of CVDs in India.

A total of 1670 anonymized medical records were collected from a tertiary hospital in South India. Seventy percent of the collected data were used to train the prediction system. Five state-of-the-art ML algorithms (k-Nearest Neighbours, Naïve Bayes, Logistic Regression, AdaBoost and Random Forest [RF]) were applied using Python programming language to develop the prediction system. The performance was evaluated over remaining 30% of data. The prediction system was later deployed in the cloud for easy accessibility via Internet.

ML effectively predicted the risk of heart disease. The best performing (RF) prediction system correctly classified 470 out of 501 medical records thus attaining a diagnostic accuracy of 93.8%. Sensitivity and specificity were observed to be 92.8% and 94.6%, respectively. The prediction system attained positive predictive value of 94% and negative predictive value of 93.6%. The prediction model developed in this study can be accessed at http://das.southeastasia.cloudapp.azure.com/predict/

Conclusions

ML-based prediction system developed in this study performs well in early diagnosis of CVDs and can be accessed via Internet. This study offers promising results suggesting potential use of ML-based heart disease prediction system as a screening tool to diagnose heart diseases in primary healthcare centres in India, which would otherwise get undetected.

Introduction

Cardiovascular diseases (CVDs) are the foremost reason of disease burden and mortality all over the world. Approximately 30% of total deaths (17.9 million) occurred due to CVDs globally in 2016. 1 The situation is critically serious in low- and middle-income countries like India. During the past three decades, the number of deaths due to CVDs has increased significantly from 15.2% to 28.1% in India. 2 Prevalence of CVDs was observed to be as high as 54.5 million cases in 2016. 3 CVDs are often detected in advanced stages amongst the underprivileged patients. Due to various reasons, Indian public healthcare system is still not capable in effectively preventing non-communicable diseases like CVDs. Efficient healthcare in terms of affordability, accessibility and quality is still far from being within reach of many. 4 Shortage of facilities in rural areas hampers medical diagnostic and therapeutic help in the initial stage of disease. Despite the government initiatives of health insurance (which are mainly for therapeutic care only) for poor people, the major section of Indian population does not have preventive health check-up benefits. 5 All these reasons lead to delayed treatment and increase in morbidity and mortality. 6

Amalgamation of ML-based prediction system in primary healthcare centres can potentially aid hugely in the prevention of CVDs in India. Recent advancements in the field of computer science have proved that machine learning (ML) algorithms can generate huge meaningful information from the immense data generated by the healthcare sector. 7 This information can be used for the diagnosis of diseases at initial stage, which can thus aid in the prevention of diseases. ML-based tools for efficient healthcare are fetching a huge attention globally. As an example, ML-based tools have recently been successfully implemented in the fields of ophthalmology and oncology in the United States. 8 , 9 Available literature reveals the development of highly accurate prediction systems for CVDs. 10 , 11 , 12 , 13 All the research studies done for early detection of CVDs as reported in the literature so far are based on the freely available online data set provided by ML repository of University of California, Irvine. 14 This data set provides information about 76 medical attributes of 303 medical records gathered from hospitals of Western countries. Information obtained from diagnostic tests like electrocardiogram, treadmill test, fluoroscopy test etc. is available in the above-mentioned data set. However, these medical tests are neither accessible nor affordable to a major section of Indian population. 15 Thus, the prediction systems developed so far are not suitable for Indian population. Moreover, some of the primary risk factors responsible for heart diseases in India, like, obesity, lack of physical activity, physiological stress, smoking and alcohol consumption etc. have not been considered in the ML-based studies so far.

As the importance of early detection of CVD is increasingly being realized, there is a definite need of developing ML-based prediction system for CVDs specifically suitable for Indian scenario.

This study was carried out with the following objectives: a) Development of a high-performance and cost-effective ML-based heart disease prediction system using routine clinical data specifically suited for Indian population and b) Deployment of the prediction system in public cloud to ensure easy accessibility via Internet particularly beneficial for rural areas in India.

Materials and methods

Study setting.

This study is an interdisciplinary research work carried out by collaboration of data scientists and specialist doctors. The study was approved by institutional ethical committee. Members of ethical committee deemed that data privacy is ensured by using anonymized medical records of existing/retrospective cohort.

Data collection

By a random selection after applying the exclusion criteria, anonymized medical records of heart patients as well as of healthy persons were collected from a tertiary hospital in South India. Anonymization ensured data privacy as the personal details of the patients were not collected for the study.

Exclusion criteria: 1) Medical data sets corresponding to pregnant females, 2) patients reporting chronic kidney disease, severe mental illness, atrial fibrillation, 3) patients who reported the prolonged use of anti-depressants, antibiotics and medicines for asthma, tuberculosis and cancer, 4) patients who are prescribed oral corticosteroids, antipsychotic drugs and immunosuppressants and 5) patients younger than 20 years or older than 100 years.

After applying these exclusion criteria, the final data set comprised of 1670 medical records belonging to people between the age 30–79 years. Study population included 881 males and 789 females. Ethnicity of all records in this study was observed to be Asian. Eight-hundred and seventy-four records did not have hypertension, while rest 796 reported hypertension. Of 1670 records, 928 reported to consume alcohol. Eight-hundred and twenty-eight records belonged to smokers. Nine-hundred and twenty records complained of stress and anxiety in life. Of 1670 records, 893 records (53.47%) were diagnosed with CVDs and remaining 777 records (46.53%) were of healthy persons with no CVDs. The persons who visited the hospital for routine check-ups and were not diagnosed with any heart disease are referred to as healthy persons (CVD risk: low) in this study.

Risk factor attributes

People living in rural parts of the country are usually unaware of the potential risk factors of heart diseases. They usually neglect the early signs of heart disease. Since the study has been carried out especially for rural areas, the clinical attributes already known to be the potential risk factors of CVDs along with lifestyle attributes associated with heart disease were chosen for this study. These attributes include age, gender, weight, height, total cholesterol levels, smoking habits, alcohol, diabetes, hypertension, family history of CVDs, intake of healthy diet, physical activity/exercise habits and stress/anxiety in life. Body mass index (BMI) was calculated internally by the software. Table 1 represents the details of risk factors considered in this study.

Description of attributes used in the study.

Attributes	Description	Categorical/numeric
Age	Years	Numeric
Weight	Kilograms	Numeric
Height	Centimetres	Numeric
Total cholesterol levels	mg/dL	Numeric
Gender	Male/female	Categorical
Hypertension	Yes/no	Categorical
Diabetes	Yes/no	Categorical
Alcohol	Yes/no	Categorical
Smoking	Yes/no	Categorical
Exercise	Yes/no	Categorical
Stress	Yes/no	Categorical
Family history of cardiovascular disease (CVD)	Yes/no	Categorical
Healthy diet	Yes/no	Categorical
Risk of CVD	High/low	Categorical

Diagnostic procedures like treadmill test and fluoroscopy (used extensively in similar studies done so far) were not considered relevant for this study to ensure cost-effectiveness. Tests for triglycerides, serum creatinine, C-reactive protein, serum fibrinogen, gamma glutamyl transferase, lipoprotein, apolipoprotein B, homocysteine, insulin test etc. although associated with the risk of heart diseases, were also not considered in this study as these medical tests are not feasible/affordable for rural population for which the research is aimed for.

Study population characteristics

Out of 1670 records, 893 were positive cases of CVD while remaining 777 records were negative cases of CVD ensuring that the data set is balanced and is not skewed in favour of any class. In the data set, mean age of patients with heart disease is 66.2 years while mean age of healthy people was 57.3 years. Mean total cholesterol for healthy people was 188 mg/dL while the mean total cholesterol for heart patients was high at 237.7 mg/dL. Mean weight of heart patients was observed to be 85.4 kg while the mean weight of healthy people was 69.4 kg. It was observed that only 27.3% of heart patients were females. Nearly 95% of healthy people reported that they used to exercise regularly. Chi-square test of independence and t -test were carried out in the study subjects on ‘prior basis’ to determine the statistical significance of categorical and numeric input attributes, respectively, in determining heart disease. 16 These tests were used to ensure the validity of data of study variables, since the performance of AI algorithms is affected by the data of variables used to train the algorithms. Descriptive characteristics of these study population variables have been represented in Table 2 .

Study Population Descriptive Characteristics.

		Total records (n = 1670)
Risk factor attribute	Unit	Cardiovascular disease (CVD) (n = 893)	No CVD (n = 777)	P-value
Age	Years (SD)	66.2 (11.2)	57.3 (12.4)	<0.001
Weight	Kilograms (SD)	85.4 (9.2)	69.4 (10.1)	<0.001
Height	Centimeters (SD)	165.7 (9.1)	162.3 (13.4)	0.23
Total cholesterol levels	mg/dL (SD)	267.7 (14.1)	218.4 (13.9)	<0.001
Gender (female)	N (%)	244 (27.3)	545 (70.1)	<0.001
Hypertension (yes)	N (%)	614 (68.7)	182 (23.4)	<0.001
Diabetes (yes)	N (%)	630 (70.5)	318 (40.9)	<0.001
Alcohol (yes)	N (%)	623 (69.7)	305 (39.2)	<0.001
Smoking (yes)	N (%)	570 (63.8)	258 (33.2)	<0.001
Exercise (yes)	N (%)	412 (46)	737 (94.8)	<0.001
Stress (yes)	N (%)	568 (63.6)	352 (45.3)	<0.001
Family history of CVD (yes)	N (%)	592 (66.2)	299 (38.4)	<0.001
Healthy diet (yes)	N (%)	496 (55.5)	398 (51.2)	0.077

∗p-value < 0.05 is statistically significant.

Methodology

Python 3.7 programming language was used for building ML-based heart disease prediction system. Powerful software libraries supported by Python namely NumPy, Pandas, Seaborn, Statsmodels.api, SciPy and Sklearn etc. were used for exploratory analysis of data 17 and implementing five ML algorithms namely k-Nearest Neighbours (k-NN), Naïve Bayes (NB), Logistic Regression (LR), AdaBoost (AB) and Random Forest (RF). This study also has typical binary classification where 13 input attributes are observed to determine if there is a high risk of heart disease in a patient (risk of CVD = high) or not (risk of CVD = low). Fig. 1 shows the workflow diagram of complete project.

Workflow diagram of the study. This figure depicts the complete workflow of the study. The medical data set of 1670 records were gathered (in random fashion). Seventy percent data samples used to train the models. Test subset comprised the rest 30% of medical records. Five machine learning algorithms are applied to train the training subset. The prediction system was hosted on the public cloud for easy accessibility.

Data pre-processing

It was observed that there were no missing values or outliers in the data.

Since the ML algorithms can process only numerical data, the categorical attributes were label encoded. Gender female was encoded as 1 while male as 0. For all the other categorical variables like diabetes, stress, exercise etc., the presence (yes) was encoded as 1 while absence (no) was encoded as 0. High risk of CVD was encoded as 1 while low risk of CVD was encoded as 0.

Building the model

Using the train_test_split function supported by scikit learn library, the complete medical data set was randomly split into two portions in the ratio 70:30 referred as training and test/validation subset, respectively. Out of total 1670 records, training subset had 1169 records while test subset had 501 records. Detailed information about the training and test subsets is provided in Table 3 . The total number of records in the training data set were 1169, of which 656 records correspond to CVDs while 513 records belonged to healthy people not diagnosed with CVDs.

Details of training and test subsets.

Class	Training subset (70%)	Test subset (30%)	Total records
High-risk cardiovascular disease (CVD)	656	237	893
Low-risk CVD	513	264	777
Total records	1169	501	1670

ML algorithms with well demonstrated performance for classification namely NB, LR and k-NN were applied to build the prediction model.

Applying ensembling algorithms for better performance

Research has proved that the performance of a ML-based prediction system can be improvised using ensembling techniques. 18 Ensembling is a union of individual classifying algorithms. Bagging ensemble algorithms namely RF and boosting ensemble algorithms namely adaptive boosting AB were also implemented for enhanced performance.

Testing the performance of the model

The performance of prediction models developed using k-NN, NB and LR algorithms was analysed using the validation subset of 501 records as shown in Table 4 . Of these records, 237 were confirmed cases of CVDs while remaining 264 records correspond to healthy people not diagnosed with CVDs. Prevalence of disease in validation subset was 237/501 = 47.3%

Performance of machine learning algorithms on validation set of 501 records.

Algorithm	True positive	True negative	False negative	False positive	Sensitivity	Specificity	Positive predictive value	Negative predictive value	Accuracy
k-Nearest Neighbours	211	230	26	34	89%	87.1%	86.1%	89.8%	88%
Naïve Bayes	210	232	27	32	88.6%	87.8%	86.7%	89.5%	88.2%
Logistic Regression	215	240	22	24	90.7%	90.9%	89.9%	91.6%	90.8%
AdaBoost	218	246	19	18	91.9%	93.1%	92.3%	92.8%	92.6%
Random Forest	220	250	17	14	92.8%	94.6%	94%	93.6%	93.8%

Analysis of confusion matrix is a standard way to check the performance of ML-based prediction system. Confusion matrix has four components namely true positives (TPs), true negatives (TNs), false positives (FPs) and false negatives (FNs).

TPs: Heart patients who are predicted correctly to have heart diseases.

TNs: Healthy persons who are predicted correctly to be healthy.

FPs: Healthy persons predicted incorrectly to have heart diseases (Type 1 error).

FNs: Heart patient predicted incorrectly to be healthy (Type 2 error).

These values are used to calculate accuracy, specificity, sensitivity, positive predictive value (PPV) and negative predictive value (NPV). PPV and NPV depend on the prevalence of disease.

A brief description of these parameters is given below.

i. Classification accuracy: This parameter represents that part of total predictions that were correct. Accuracy = (TN + TP)/(TN + FN + FP + TP)
ii. Sensitivity: This parameter reflects the ratio of cases that were accurately predicted with heart disease to the total number of actual cases of heart disease. Mathematically, sensitivity = TP/TP + FN
iii. Specificity: This parameter calculates the ratio of cases that are correctly predicted with no heart disease to the entire count of actual cases with no heart disease. Mathematically, Specificity = TN/FP + TN
iv. PPV: This parameter reflects the ratio of cases that are correctly predicted with heart diseases to the total count of cases predicted to have heart disease. Mathematically, PPV = TP/TP + FP
v. NPV: This parameter reflects the ratio of cases correctly predicted to be healthy to the total count of cases predicted to be healthy. Mathematically, NPV = TN/TN + FN

Fine tuning of hyperparameters

Grid Search for cross-validation was used to identify the best hyperparameters for the learning algorithms. Grid Search CV class from sklearn library was used for this purpose.

Deployment on the public cloud

The best performance prediction system built using RF model was deployed in Microsoft Azure cloud for better accessibility. 19 ‘ Pickle ’ and ‘ Flask ’ software libraries of Python programming language were used for this purpose. 20 Hosting the prediction system on cloud enables it to be easily accessed from anywhere in the world via Internet. This is highly useful feature for healthcare sector of India, which faces the major issue of shortage of medical facilities especially in rural areas. Accessing this prediction system is as easy as accessing an e-mail via Internet.

CVD prediction system was developed by applying five well-established ML algorithms on the training data set. The performance was tested on the validation test set of 501 records. Prevalence of disease in validation subset was 237/501 = 47.3% Performance metrics namely accuracy, sensitivity, specificity, PPV and NPV were calculated for each algorithm. The performance results of all classifiers are given in Table 4 .

The best hyperparameters for k-NN (n_neighbors = 12) resulted in a performance of sensitivity 89%, specificity 87.1%, PPV 86.1%, NPV 89.8%. The performance of NB was found to better than k-NN. Sensitivity 88.6%, specificity 87.8%, PPV 86.7%, NPV 89.5% were achieved by NB.

LR with hyperparameters (C = 1, penalty = l2) performed well in classifying people with low risk or high risk of CVDs. LR correctly classified 455 out of 501 records, thus attaining a classification accuracy of 90.8%. Sensitivity 90.7% and specificity were 90.7% and 90.9%, respectively. PPV was observed to be 89.9% while NPV was 91.6%.

Models built using ensemble techniques (RF and AB) performed better than LR. AB model was trained with Stage-wise Adaptive Modelling using a Multi-class Exponential loss function (n_estimators = 30) while RF based on ‘gini index’ with n_estimators = 150 resulted in the best performance. Sensitivity and specificity of AB model was 91.9% and 93.1%, respectively, while RF reported 92.8% sensitivity and 94.6% specificity. PPV 94% and NPV 93.6% were achieved by RF–based prediction model.

Interpretation of ML-based models is not easy, and these are usually considered as ‘black boxes. However, logistic regression–based models are quite interpretable. Logistic regression was implemented using the Logit function (Binomial family) based on maximum likelihood estimation method to predict CVD risk using statsmodels.api library of Python. Fig. 2 shows the summary of results obtained.

Study population characteristics mean (standard deviation) of numerical attributes along with p-values of t -test to indicate the statistical significance for two groups: high risk/low risk of cardiovascular disease (CVDs). Count (%) of categorical attributes in two groups: high risk/low risk of CVDs.

Male gender, diabetes, hypertension, high cholesterol level, smoking and alcohol were significantly associated with CVD. Lack of exercise and stress were observed to be more prevalent in CVD group (p value < 0.05).

Estimate column in the summary reflects the natural logarithm of odds ratio of getting diagnosed with high risk of heart disease keeping all other features constant. Due to negative values of log (odds ratio) it is inferred that females had a low risk of CVDs compared with males. Regular exercise and intake of healthy diet were observed to be associated with low risk of CVDs; on the other hand, diabetes, hypertension, stress, smoking and family history tend to result in high risk of CVDs.

The odds ratio column in the summary suggests how the odds ratio of being detected with high risk of CVD change if all other attributes are kept constant. Hypertension tends to increase the odds ratio of high risk of CVDs by 1.573 while the odds ratio drops significantly to 0.328 with regular physical exercise. Odds ratio of high risk of CVD for females is 0.788 compared with males.

Ensemble algorithms (RF and AB) are based on decision trees and attribute importance is graded according to selection occurrence frequency of an attribute as a decision node decided based on information gain and entropy. Variable importance for boosting algorithm was decided based on the impurity-based scores using feature_importances_ from sklearn library of Python. Attributes exercise, weight, total cholesterol, hypertension and age were the top five important attributes for AB algorithm. In case of RF prediction system, variable importance scores for attributes weight, exercise, total cholesterol, hypertension, and gender were found to be maximum for predicting CVDs. Variable importance for AB algorithms and RF is represented graphically in Fig. 3 (a) and (b), respectively.

Variable importance. (a) Variable importance for AdaBoost-based prediction model. (b) Variable importance for Random Forest–based prediction model.

RF-based CVD prediction model (trained on 1169 records and tested on 501 records) is hosted on cloud and can be easily accessed at das.southeastasia.cloudapp.azure.com/predict/

The input attributes of the patient are entered into the system. The system predicts if the patient has low risk of CVDs or high risk. Sample screenshots of the result obtained using the prediction system are shown in Fig. 4 .

Using cardiovascular disease (CVD) prediction model to test the risk of CVDs. The medical practitioner enters the patient's clinical parameters as well as attributes related to his lifestyle to predict the risk of CVD.

In the recent years, substantial research studies have been carried out to build methods for diagnosing heart diseases in early stages. Various feature selection techniques were applied in the research carried out by Takci. 21 (2018), and the resulting prediction system attained an accuracy of 84.81%. Similar study was carried out by Kausar et al. and an accuracy of 88.41% was obtained. 22 Prediction system developed by Khalid Raza using ensembling technique (2019) attained an accuracy of 88.88%. 23 A similar accuracy level of 89% was achieved by the prediction system developed by Haq et al. in 2019. 24 Using artificial neural network to design a prediction system Alic et al. achieved an accuracy of 91% in their research study. 25 But importantly, the prediction system developed in all of these studies do not work effectively well for Indian population as these models are based on data collected from Western countries and do not take into consideration lifestyle-related risk factors responsible for CVDs (lack of physical activity, family history, alcohol etc.). Moreover, these systems rely on the results of medical tests like ECG, treadmill test, fluoroscopy tests etc., which are not feasible in Indian primary health centres in the existing scenario.

The accuracy attained in the present study is 93.8%. The prediction system developed in this research uses 13 clinical parameters and identifies the risk of a person to have heart disease. Compared with the studies done so far, this study has been carried out on Indian population, and the potential risk factors like high body weight, lack of exercise, psychological stress, family history, smoking and alcohol consumption habits have been considered in this study (unlike the studies quoted previously). It is worth noting that the system developed in this study is highly cost-effective compared with earlier studies as expensive tests like fluoroscopy and treadmill tests have not been taken into consideration. Easy accessibility of the prediction system via Internet is also an added remarkable feature of this study, which was not reported by earlier studies. It is worth mentioning here that prediction model developed in this pilot study predicts output depending on the study population attribute trends it was trained on. Once the ML models are trained and tested on voluminous data sets, it can be used as a screening tool in rural India and can help in the prevention of CVDs.

Cost-effectiveness, excellent performance and easy accessibility of the prediction system via Internet defend the use of ML-based prediction system as a screening tool for CVD detection in India.

To the best of our knowledge, this study was first of its kind in Indian context. Developed countries like the United Kingdom and the United States are investing their resources to carry to research for developing ML-based prediction models for diagnosing heart diseases in primary healthcare centres. 26 , 27 It is recommended that similar studies should be promoted in India. The current national health policy (2017) of our government, laying stress on preventive health will be more meaningful and fruitful if advancement in this field is made as early as possible. 28 We propose larger studies of multicentric nature for development of AI prediction systems for CVD screening in our country, which is facing ever increasing load of morbidity and mortality due to CVD being detected in late advanced stages. Premier institutes of medicine and technology can collaborate in this regard to diagnose other lifestyle diseases and non-communicable diseases like malignancies. Cardiological Society of India (CSI) can help in this regard. Other modern techniques like artificial neural networks can be applied to further improve the performance of the system.

Limitations of the study

This study used a data set of 1670 patients reporting to a tertiary care private setup in a south Indian metropolitan city where largely the higher income group seeks the medical care. This potentially may seem biased in reader's mind, but this study was aimed only to detect the robustness of a prediction model based on ML. The results obtained from the prediction system developed in this study are based on the attribute trends of the study population on which the model is trained on. In future the model needs to be trained on huge data sets collected from diverse regions before using it as a screening tool.

The study portrays the capability of ML algorithms to predict CVDs in Indian population. Issues of affordability and accessibility in healthcare sector of India can be addressed using ML-based models, which can be easily accessed via Internet even in the rural parts of the country. It is proposed to build and test the performance of similar systems using voluminous cardiac data sets belonging to all economic sections of the society collected from various regions of India. We recommend similar studies of multicentric nature across entire country. To achieve the sustainable development goals laid down by World Health Organization, it is high time, we as a country do take timely advantage of ML-based prediction systems in improving preventive care aspect of public healthcare system. 29

What is already known?

ML-based tools have shown remarkable performance in diagnosing various serious diseases in initial stages in healthcare centres of developed countries.

What does this study add?

An indigenous high-performance ML-based CVD prediction system easily accessible via Internet is proposed for existing Indian healthcare system. Healthcare in India can be made more affordable and accessible using ML-based prediction systems.

Disclosure of competing interest

The authors have none to declare.

Acknowledgements

The authors express their heartfelt gratitude to Sagar Hospitals, Jayanagar, Bengaluru, for providing anonymized information of patients' health parameters for carrying out this study. No funding was received for this project.

Thank you for visiting nature.com. You are using a browser version with limited support for CSS. To obtain the best experience, we recommend you use a more up to date browser (or turn off compatibility mode in Internet Explorer). In the meantime, to ensure continued support, we are displaying the site without styles and JavaScript.

View all journals
Explore content
About the journal
Publish with us
Sign up for alerts
Published: 02 September 2024

An orally administered glucose-responsive polymeric complex for high-efficiency and safe delivery of insulin in mice and pigs

Kangfan Ji 1 , 2 ,
Xiangqian Wei 1 , 2 ,
Anna R. Kahkoska 3 ,
Juan Zhang 1 , 2 ,
Yang Zhang 1 , 2 ,
Jianchang Xu 1 , 2 ,
Xinwei Wei 1 , 2 ,
Wei Liu 1 , 2 ,
Yanfang Wang 1 , 2 ,
Yuejun Yao 1 , 2 ,
Xuehui Huang 1 , 2 ,
Shaoqian Mei 1 , 2 ,
Yun Liu 1 , 2 ,
Shiqi Wang 1 , 2 ,
Zhengjie Zhao 1 , 2 ,
Ziyi Lu 1 , 2 ,
Jiahuan You 1 , 2 ,
Guangzheng Xu 1 , 2 ,
Youqing Shen ORCID: orcid.org/0000-0003-1837-7976 4 ,
John. B. Buse ORCID: orcid.org/0000-0002-9723-3876 5 ,
Jinqiang Wang ORCID: orcid.org/0000-0002-0048-838X 1 , 2 , 6 , 7 &
Zhen Gu ORCID: orcid.org/0000-0003-2947-4456 1 , 2 , 6 , 8 , 9 , 10 , 11

Nature Nanotechnology ( 2024 ) Cite this article

6 Altmetric

Metrics details

Engineering
Nanoscience and technology

Contrary to current insulin formulations, endogenous insulin has direct access to the portal vein, regulating glucose metabolism in the liver with minimal hypoglycaemia. Here we report the synthesis of an amphiphilic diblock copolymer comprising a glucose-responsive positively charged segment and polycarboxybetaine. The mixing of this polymer with insulin facilitates the formation of worm-like micelles, achieving highly efficient absorption by the gastrointestinal tract and the creation of a glucose-responsive reservoir in the liver. Under hyperglycaemic conditions, the polymer triggers a rapid release of insulin, establishing a portal-to-peripheral insulin gradient—similarly to endogenous insulin—for the safe regulation of blood glucose. This insulin formulation exhibits a dose-dependent blood-glucose-regulating effect in a streptozotocin-induced mouse model of type 1 diabetes and controls the blood glucose at normoglycaemia for one day in non-obese diabetic mice. In addition, the formulation demonstrates a blood-glucose-lowering effect for one day in a pig model of type 1 diabetes without observable hypoglycaemia, showing promise for the safe and effective management of type 1 diabetes.

This is a preview of subscription content, access via your institution

Access options

Access Nature and 54 other Nature Portfolio journals

Get Nature+, our best-value online-access subscription

24,99 € / 30 days

cancel any time

Subscribe to this journal

Receive 12 print issues and online access

251,40 € per year

only 20,95 € per issue

Buy this article

Purchase on SpringerLink
Instant access to full article PDF

Prices may be subject to local taxes which are calculated during checkout

Week-long normoglycaemia in diabetic mice and minipigs via a subcutaneous dose of a glucose-responsive insulin complex

Metabolic and immunomodulatory control of type 1 diabetes via orally delivered bile-acid-polymer nanocarriers of insulin or rapamycin

A co-formulation of supramolecularly stabilized insulin and pramlintide enhances mealtime glucagon suppression in diabetic pigs

Data availability.

The data that support the results of this study are available within the Article and its Supplementary Information. Source data are provided with this paper.

Sun, H. et al. IDF Diabetes Atlas: Global, regional and country-level diabetes prevalence estimates for 2021 and projections for 2045. Diabetes Res. Clin. Pract. 183 , 109119 (2022).

Article PubMed Google Scholar

Katsarou, A. et al. Type 1 diabetes mellitus. Nat. Rev. Dis. Primers 3 , 17016 (2017).

Mathieu, C., Martens, P. J. & Vangoitsenhoven, R. One hundred years of insulin therapy. Nat. Rev. Endocrinol. 17 , 715–725 (2021).

Mitragotri, S., Burke, P. A. & Langer, R. Overcoming the challenges in administering biopharmaceuticals: formulation and delivery strategies. Nat. Rev. Drug Discov. 13 , 655–672 (2014).

Article CAS PubMed PubMed Central Google Scholar

Geho, W. B. The importance of the liver in insulin replacement therapy in insulin-deficient diabetes. Diabetes 63 , 1445–1447 (2014).

Article CAS PubMed Google Scholar

Brown, T. D., Whitehead, K. A. & Mitragotri, S. Materials for oral delivery of proteins and peptides. Nat. Rev. Mater. 5 , 127–148 (2020).

Article Google Scholar

Lee, J. S. et al. Metabolic and immunomodulatory control of type 1 diabetes via orally delivered bile-acid-polymer nanocarriers of insulin or rapamycin. Nat. Biomed. Eng. 5 , 983–997 (2021).

Veiseh, O., Tang, B. C., Whitehead, K. A., Anderson, D. G. & Langer, R. Managing diabetes with nanomedicine: challenges and opportunities. Nat. Rev. Drug Discov. 14 , 45–57 (2015).

Lu, Y., Aimetti, A. A., Langer, R. & Gu, Z. Bioresponsive materials. Nat. Rev. Mater. 2 , 16075 (2016).

Bakh, N. A. et al. Glucose-responsive insulin by molecular and physical design. Nat. Chem. 9 , 937–944 (2017).

Wang, J. et al. Glucose-responsive insulin and delivery systems: innovation and translation. Adv. Mater. 32 , 1902004 (2020).

Article CAS Google Scholar

Mo, R., Jiang, T., Di, J., Tai, W. & Gu, Z. Emerging micro- and nanotechnology based synthetic approaches for insulin delivery. Chem. Soc. Rev. 43 , 3595–3629 (2014).

Wang, Z., Wang, J., Kahkoska, A. R., Buse, J. B. & Gu, Z. Developing insulin delivery devices with glucose responsiveness. Trends Pharmacol. Sci. 42 , 31–44 (2021).

Gu, Z. et al. Glucose-responsive microgels integrated with enzyme nanocapsules for closed-loop insulin delivery. ACS Nano 7 , 6758–6766 (2013).

Chen, Z. et al. Synthetic beta cells for fusion-mediated dynamic insulin secretion. Nat. Chem. Biol. 14 , 86–93 (2018).

Podual, K., Doyle, F. J. & Peppas, N. A. Glucose-sensitivity of glucose oxidase-containing cationic copolymer hydrogels having poly(ethylene glycol) grafts. J. Control. Release 67 , 9–17 (2000).

Kitano, S., Koyama, Y., Kataoka, K., Okano, T. & Sakurai, Y. A novel drug delivery system utilizing a glucose responsive polymer complex between poly (vinyl alcohol) and poly ( N -vinyl-2-pyrrolidone) with a phenylboronic acid moiety. J. Control. Release 19 , 161–170 (1992).

Hisamitsu, I., Kataoka, K., Okano, T. & Sakurai, Y. Glucose-responsive gel from phenylborate polymer and poly (vinyl alcohol): prompt response at physiological pH through the interaction of borate with amino group in the gel. Pharm. Res. 14 , 289–293 (1997).

Matsumoto, A. et al. A synthetic approach toward a self-regulated insulin delivery system. Angew. Chem. Int. Ed. 51 , 2124–2128 (2012).

Chou, D. H.-C. et al. Glucose-responsive insulin activity by covalent modification with aliphatic phenylboronic acid conjugates. Proc. Natl Acad. Sci. USA 112 , 2401–2406 (2015).

Wang, J. et al. Charge-switchable polymeric complex for glucose-responsive insulin delivery in mice and pigs. Sci. Adv. 5 , eaaw4357 (2019).

Yu, J. et al. Glucose-responsive insulin patch for the regulation of blood glucose in mice and minipigs. Nat. Biomed. Eng. 4 , 499–506 (2020).

Wang, J. et al. Injectable biodegradable polymeric complex for glucose-responsive insulin delivery. ACS Nano 15 , 4294–4304 (2021).

Matsumoto, A. et al. Synthetic ‘smart gel’ provides glucose-responsive insulin delivery in diabetic mice. Sci. Adv. 3 , eaaq0723 (2017).

Article PubMed PubMed Central Google Scholar

Zhang, J. et al. Week-long normoglycaemia in diabetic mice and minipigs via a subcutaneous dose of a glucose-responsive insulin complex. Nat. Biomed. Eng. https://doi.org/10.1038/s41551-023-01138-7 (2023).

Brownlee, M. & Cerami, A. A glucose-controlled insulin-delivery system: semisynthetic insulin bound to lectin. Science 206 , 1190–1191 (1979).

Wang, J. et al. Glucose transporter inhibitor-conjugated insulin mitigates hypoglycemia. Proc. Natl Acad. Sci. USA 116 , 10744–10748 (2019).

Yao, Y., Ji, K., Wang, Y., Gu, Z. & Wang, J. Materials and carriers development for glucose-responsive insulin. Acc. Mater. Res. 3 , 960–970 (2022).

Chu, J. N. & Traverso, G. Foundations of gastrointestinal-based drug delivery and future developments. Nat. Rev. Gastroenterol. Hepatol. 19 , 219–238 (2022).

Yang, T. et al. Ligand-switchable nanoparticles resembling viral surface for sequential drug delivery and improved oral insulin therapy. Nat. Commun. 13 , 6649 (2022).

Xi, Z. et al. Dual-modified nanoparticles overcome sequential absorption barriers for oral insulin delivery. J. Control. Release 342 , 1–13 (2022).

Drucker, D. J. Advances in oral peptide therapeutics. Nat. Rev. Drug Discov. 19 , 277–289 (2020).

Yang, Y. et al. Recent advances in oral and transdermal protein delivery systems. Angew. Chem. Int. Ed. 62 , e202214795 (2023).

Baryakova, T. H., Pogostin, B. H., Langer, R. & McHugh, K. J. Overcoming barriers to patient adherence: the case for developing innovative drug delivery systems. Nat. Rev. Drug Discov. 22 , 387–409 (2023).

Lamson, N. G., Berger, A., Fein, K. C. & Whitehead, K. A. Anionic nanoparticles enable the oral delivery of proteins by enhancing intestinal permeability. Nat. Biomed. Eng. 4 , 84–96 (2020).

Lagarrigue, P., Moncalvo, F. & Cellesi, F. Non-spherical polymeric nanocarriers for therapeutics: the effect of shape on biological systems and drug delivery properties. Pharmaceutics 15 , 32 (2023).

Ji, K. et al. Material design for oral insulin delivery. Med X 1 , 7 (2023).

PubMed PubMed Central Google Scholar

Yu, J. et al. Glucose-responsive oral insulin delivery for postprandial glycemic regulation. Nano Res. 12 , 1539–1545 (2019).

Wang, A. et al. Liver-target and glucose-responsive polymersomes toward mimicking endogenous insulin secretion with improved hepatic glucose utilization. Adv. Funct. Mater. 30 , 1910168 (2020).

Xiao, Y. et al. Glucose-responsive oral insulin delivery platform for one treatment a day in diabetes. Matter 4 , 3269–3285 (2021).

Han, X. et al. Zwitterionic micelles efficiently deliver oral insulin without opening tight junctions. Nat. Nanotechnol. 15 , 605–614 (2020).

Lu, Y. et al. Micelles with ultralow critical micelle concentration as carriers for drug delivery. Nat. Biomed. Eng. 2 , 318–325 (2018).

Banerjee, A., Qi, J., Gogoi, R., Wong, J. & Mitragotri, S. Role of nanoparticle size, shape and surface chemistry in oral drug delivery. J. Control. Release 238 , 176–185 (2016).

Xiao, W. et al. Highly sensitive colorimetric detection of a variety of analytes via the Tyndall effect. Anal. Chem. 91 , 15114–15122 (2019).

Cone, R. A. Barrier properties of mucus. Adv. Drug Deliv. Rev. 61 , 75–85 (2009).

Pridgen, E. M. et al. Transepithelial transport of Fc-targeted nanoparticles by the neonatal Fc receptor for oral delivery. Sci. Transl. Med. 5 , 213ra167 (2013).

Acosta-Rodríguez, V. et al. Circadian alignment of early onset caloric restriction promotes longevity in male c57BL/6J mice. Science 376 , 1192–1202 (2022).

Yang, R. et al. A glucose-responsive insulin therapy protects animals against hypoglycemia. JCI Insight 3 , e97476 (2018).

Qiu, Y. et al. Long-lasting designer insulin with glucose-dependent solubility markedly reduces risk of hypoglycemia. Adv. Ther. (Weinh.) 2 , 1900128 (2019).

Ayala, J. E. et al. Hyperinsulinemic–euglycemic clamps in conscious, unrestrained mice. J. Vis. Exp. 16 , e3188 (2011).

Google Scholar

Cao, Z., Zhang, L. & Jiang, S. Superhydrophilic zwitterionic polymers stabilize liposomes. Langmuir 28 , 11625–11632 (2012).

Fan, W. et al. Mucus penetrating and cell-binding polyzwitterionic micelles as potent oral nanomedicine for cancer drug delivery. Adv. Mater. 34 , 2109189 (2022).

Zou, J.-J. et al. Efficient oral insulin delivery enabled by transferrin-coated acid-resistant metal–organic framework nanoparticles. Sci. Adv. 8 , eabm4677 (2022).

Download references

Acknowledgements

This work was supported by grants from the National Key R&D Program of China (2022YFE0202200, J.W.), JDRF (2-SRA-2021-1064-M-B, Z.G.; 2-SRA-2022-1159-M-B, J.W.), the Key Project of Science and Technology Commission of Zhejiang Province (2024C03083, Z.G.; 2024C03085, J.W.), Zhejiang University’s start-up packages and the Starry Night Science Fund at Shanghai institute for Advanced Study of Zhejiang University (SN-ZJU-SIAS-009, J.W.). A.R.K. is supported by the National Center for Advancing Translational Sciences, National Institutes of Health (KL2TR002490, J.W.). The project was supported by the Clinical and Translational Science Award program of the National Center for Advancing Translational Science, National Institutes of Health (UL1TR002489, J.W.). We appreciate the help from J. Pan and D. Wu of the Research and Service Center (College of Pharmaceutical Science, Zhejiang University) for technical support, G. Z. and Y. Zhang (Cryo-EM centre, Zhejiang University) for processing the samples for electron microscopy and D. Xu, M. Zhang, S. Xiong and D. Chen (Disease Simulation and Animal Model Platform of Liangzhu Laboratory) for taking care of the minipigs.

Author information

Authors and affiliations.

State Key Laboratory of Advanced Drug Delivery and Release Systems, College of Pharmaceutical Sciences, Zhejiang University, Hangzhou, China

Kangfan Ji, Xiangqian Wei, Juan Zhang, Yang Zhang, Jianchang Xu, Xinwei Wei, Wei Liu, Yanfang Wang, Yuejun Yao, Xuehui Huang, Shaoqian Mei, Yun Liu, Shiqi Wang, Zhengjie Zhao, Ziyi Lu, Jiahuan You, Guangzheng Xu, Jinqiang Wang & Zhen Gu

Jinhua Institute of Zhejiang University, Jinhua, China

Department of Nutrition, University of North Carolina at Chapel Hill, Chapel Hill, NC, USA

Anna R. Kahkoska

Zhejiang Key Laboratory of Smart Biomaterials, College of Chemical and Biological Engineering, Zhejiang University, Hangzhou, China

Youqing Shen

Department of Medicine, University of North Carolina School of Medicine, Chapel Hill, NC, USA

John. B. Buse

Key Laboratory of Advanced Drug Delivery Systems of Zhejiang Province, College of Pharmaceutical Sciences, Zhejiang University, Hangzhou, China

Jinqiang Wang & Zhen Gu

Department of Pharmacy, Second Affiliated Hospital, Zhejiang University School of Medicine, Zhejiang University, Hangzhou, China

Jinqiang Wang

Department of General Surgery, Sir Run Run Shaw Hospital, School of Medicine, Zhejiang University, Hangzhou, China

Liangzhu Laboratory, Hangzhou, China

Institute of Fundamental and Transdisciplinary Research, Zhejiang University, Hangzhou, China

MOE Key Laboratory of Macromolecular Synthesis and Functionalization, Department of Polymer Science and Engineering, Zhejiang University, Hangzhou, China

You can also search for this author in PubMed Google Scholar

Contributions

Z.G., J.W., Y.S. and J.B.B. conceived and designed the study. K.J., Xiangqian Wei, J.Z., J.X., Xinwei Wei, Y.Z., W.L., Y.W., Y.Y., S.M. and Y.L. conducted experiments and obtained related data. X.H., S.W., Z.Z., J.Y., G.X. and Z.L. gave experimental operation and theoretical guidance of mice experiments. K.J., Xiangqian Wei, J.Z. and J.X. conducted minipigs experiments and provided theoretical support. Z.G., J.W., Y.S., K.J., J.Z., Xiangqian Wei, A.R.K., J.B.B. and J.X. analysed the data and wrote the paper.

Corresponding authors

Correspondence to Jinqiang Wang or Zhen Gu .

Ethics declarations

Competing interests.

Z.G. is the co-founder of Zenomics Inc., Zcapsule Inc. and μ Zen Inc. The other authors declare no competing interests.

Peer review

Peer review information.

Nature Nanotechnology thanks Kåre Birkeland and Nicholas Hunt for their contribution to the peer review of this work.

Additional information

Publisher’s note Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations.

Extended data

Extended data fig. 1 bg-regulating effects in diabetic minipigs..

BG of diabetic minipigs treated with the insulin capsules (oral), the PPF-ins capsules (oral) or Lantus (s.c.). The insulin dose of oral formulations was set to 4.2 U/kg. The Lantus dose was set to 0.3 U/kg.

Source data

Supplementary information, supplementary information.

Supplementary Figs. 1–24.

Reporting Summary

Supplementary data 1.

Supplementary statistical source data.

Source Data Fig. 1

Statistical source data for Fig. 1.

Source Data Fig. 2

Statistical source data for Fig. 2.

Source Data Fig. 4

Statistical source data for Fig. 4.

Source Data Fig. 5

Statistical source data for Fig. 5.

Source Data Fig. 6

Statistical source data for Fig. 6.

Source Data Extended Data Fig. 1

Statistical source data for Extended Data Fig. 1.

Rights and permissions

Springer Nature or its licensor (e.g. a society or other partner) holds exclusive rights to this article under a publishing agreement with the author(s) or other rightsholder(s); author self-archiving of the accepted manuscript version of this article is solely governed by the terms of such publishing agreement and applicable law.

Reprints and permissions

About this article

Cite this article.

Ji, K., Wei, X., Kahkoska, A.R. et al. An orally administered glucose-responsive polymeric complex for high-efficiency and safe delivery of insulin in mice and pigs. Nat. Nanotechnol. (2024). https://doi.org/10.1038/s41565-024-01764-5

Download citation

Received : 12 August 2023

Accepted : 19 July 2024

Published : 02 September 2024

DOI : https://doi.org/10.1038/s41565-024-01764-5

Share this article

Anyone you share the following link with will be able to read this content:

Sorry, a shareable link is not currently available for this article.

Provided by the Springer Nature SharedIt content-sharing initiative

Quick links

Explore articles by subject
Guide to authors
Editorial policies

IMAGES

Most Important Outcomes Research Papers on Cardiovascular Disease in
(PDF) Heart Disease Prediction using Machine Learning Techniques
(PDF) Heart disease prediction using machine learning techniques: A survey
(PDF) Heart Disease Prediction using Machine Learning Techniques
(PDF) PREDICTION OF HEART DISEASE BY USING MACHINE LEARNING
(PDF) ANALYSIS OF HEART DISEASE PREDICTION USING MACHINE LEARNING

VIDEO

Heart Disease Research

COMMENTS

Global Burden of Cardiovascular Diseases and Risk Factors, 1990-2019
Global Burden of Disease Study Methods. GBD 2019 is a multinational collaborative research study that estimates disease burden for every country in the world (1,4).The study is an ongoing effort, updated annually, and is designed to allow for consistent comparison over time from 1990 to 2019, by age and sex, and across locations.
Cardiology
Atrial Fibrillation and Stable Coronary Artery Disease. G.Y. Lip10.1056/NEJMe2409696. Patients with atrial fibrillation are at high risk for stroke, and international guidelines recommend oral ...
JAMA Cardiology
Explore the latest advances in PCI and device therapy, treatment of myocardial infarction, dyslipidemia, heart failure, afib & rhythm disorders, TAVR, and ... 1990-2017 Global Disparities in Parkinson Disease Guidelines for ... and Policy Promoting EDI in Genetics Research PTSD and Cardiovascular Disease Red Blood Cell Transfusion: 2023 AABB ...
A Systematic Review of Major Cardiovascular Risk Factors: A Growing
Introduction and background. Atherosclerosis is the process of formation of plaque composed of cholesterol, fat, calcium, and other substances in the wall of large and medium-sized arteries causing diminished blood flow to an area of the body [1,2].Moreover, atherosclerotic vascular disease can be classified into two arenas: a cardiovascular disease affecting the heart and peripheral blood ...
Early and accurate detection and diagnosis of heart disease using
Heart disease is a fatal human disease, rapidly increases globally in both developed and undeveloped countries and consequently, causes death. ... research study recommends that which feature ...
Heart Disease and Stroke Statistics—2022 Update: A Report From the
Background: The American Heart Association, in conjunction with the National Institutes of Health, annually reports the most up-to-date statistics related to heart disease, stroke, and cardiovascular risk factors, including core health behaviors (smoking, physical activity, diet, and weight) and health factors (cholesterol, blood pressure, and glucose control) that contribute to cardiovascular ...
2024 Heart Disease and Stroke Statistics: A Report of US and Global
BACKGROUND: The American Heart Association (AHA), in conjunction with the National Institutes of Health, annually reports the most up-to-date statistics related to heart disease, stroke, and cardiovascular risk factors, including core health behaviors (smoking, physical activity, nutrition, sleep, and obesity) and health factors (cholesterol, blood pressure, glucose control, and metabolic ...
Hot topics and trends in cardiovascular research
In clinical research, patient management after myocardial infarction (MI) and outside the hospital are leading topics, but the true 'hot' topic was aortic valve disease that saw a surge of interest, related to transaortic valve repair, starting 2008. Though still small in numbers, heart failure research and stem cells saw substantial growth.
Cardiovascular diseases
In this Review, Redfield and colleagues highlight the importance of studying human cardiac tissue in HFpEF, discuss sources, challenges and methods for studying human myocardial samples, summarize ...
Nature Cardiovascular Research
Aficamten is a small-molecule cardiac myosin inhibitor designed to treat hypertrophic cardiomyopathy. Hartman et al. use mouse models of cardiac function to show that aficamten decreases the ...
The Global Burden of Cardiovascular Diseases and Risk:
Introduction. Cardiovascular diseases (CVDs) have collectively remained the leading causes of death worldwide and substantially contribute to loss of health and excess health system costs. 1, 2 The Global Burden of Diseases, Injuries, and Risk Factors (GBD) Study has tracked trends in death and disability since 1990 and has provided an updated perspective on the status of cardiovascular health ...
A Brief Review of Cardiovascular Diseases, Associated Risk ...
Abstract. Cardiovascular diseases (CVDs) are the leading cause of premature death and disability in humans and their incidence is on the rise globally. Given their substantial contribution towards the escalating costs of health care, CVDs also generate a high socio-economic burden in the general population.
Heart Disease and Stroke Statistics—2020 Update: A Report From the
Background: The American Heart Association, in conjunction with the National Institutes of Health, annually reports on the most up-to-date statistics related to heart disease, stroke, and cardiovascular risk factors, including core health behaviors (smoking, physical activity, diet, and weight) and health factors (cholesterol, blood pressure, and glucose control) that contribute to ...
State of the Science: The Relevance of Symptoms in Cardiovascular
Symptoms are subjective experiences that may indicate disease or significant change in health status. Symptoms have been linked to cardiovascular disease (CVD) since Egyptian physicians and Hippocrates described fatigue and dyspnea, respectively, as being related to the failing heart. 1,2 In a contemporary view of CVD, symptoms often are critical elements of the diagnosis, evaluation ...
Cardiovascular diseases
Cardiovascular diseases articles from across Nature Portfolio. Atom; RSS Feed; Definition. Cardiovascular diseases are pathological conditions affecting the heart and/or blood vessels that is, the ...
Coronary Heart Disease Research
Heart disease, including coronary heart disease, remains the leading cause of death in the United States. However, the rate of heart disease deaths has declined by 70% over the past 50 years, thanks in part to NHLBI-funded research. Many current studies funded by the NHLBI focus on discovering genetic associations and finding new ways to ...
Cardiovascular disease (CVD) and associated risk factors among older
Ischaemic heart disease and cerebrovascular disease (stroke) combined accounted for more than 85.1% of all cardiovascular disease deaths in 2016. Our study indicated that CVDs(angina, stroke) were prevalent and variable among older adults in six countries. Angina and stroke were both highest in Russian Federation(47.5%, 6.1% respectively).
Primary prevention of cardiovascular disease: A review of contemporary
Cardiovascular disease (CVD) is an umbrella term for a number of linked pathologies, commonly defined as coronary heart disease (CHD), cerebrovascular disease, peripheral arterial disease, rheumatic and congenital heart diseases and venous thromboembolism. Globally CVD accounts for 31% of mortality, the majority of this in the form of CHD and ...
Early prediction of heart disease with data analysis using supervised
Heart diseases are consistently ranked among the top causes of mortality on a global scale. Early detection and accurate heart disease prediction can help effectively manage and prevent the disease. However, the traditional methods have failed to improve heart disease classification performance. So, this article proposes a machine learning approach for heart disease prediction (HDP) using a ...
The right heart in patients with cancer. A scientific statement of the
Interdepartmental Hypertension Research Center (CIRIAPA), 'Federico II' University, Naples, Italy. Center for Basic and Clinical Immunology Research (CISI), 'Federico II' University, Naples, Italy. Search for more papers by this author
Heart health
The science of heart health focuses on several related biological phenomena, including inflammation, cholesterol level and a lingering form of immune response that tends to exacerbate the risk of ...
Coronary Artery Disease
References. Since its inception, articles published in Arteriosclerosis, Thrombosis, and Vascular Biology (ATVB) have contributed to our understanding of coronary artery disease (CAD) and its different complex pathophysiological processes. Here, we review articles related to CAD published in ATVB in the past 2 years from 2018 to 2019.
Signs of heart disease may be more subtle in women than men
In a review paper, the American Heart Association spells out how symptoms of heart attack, stroke and other cardiovascular problems differ in women and men. Aug. 22, 2022, 4:09 PM UTC / Source ...
Machine learning-based heart disease prediction system for Indian
Introduction. Cardiovascular diseases (CVDs) are the foremost reason of disease burden and mortality all over the world. Approximately 30% of total deaths (17.9 million) occurred due to CVDs globally in 2016. 1 The situation is critically serious in low- and middle-income countries like India. During the past three decades, the number of deaths due to CVDs has increased significantly from 15.2 ...
Coronary Artery Disease and Its Risk Factors
Coronary artery disease (CAD) is the world-wide leading cause of death not only in high-income countries but also increasingly in developing countries. 1 Although death rates from CAD have decreased in most high- and middle-income countries in the past 2 decades, there are worrying signs of a lessening trend in the United States, 2 and the dramatic increases of world-wide obesity 3 and ...
An orally administered glucose-responsive polymeric complex for high
The major organs included heart, lung, liver, spleen, kidney (from up to down, left) and intestine (right). ... (UL1TR002489, J.W.). We appreciate the help from J. Pan and D. Wu of the Research ...

Just Published

Save citation to file

Add to My Bibliography

A Brief Review of Cardiovascular Diseases, Associated Risk Factors and Current Treatment Regimes

Similar articles

Publication types

Related information

Coronary Heart Disease Research

NHLBI research that really made a difference

Current research funded by the NHLBI

Research funding

Current research on preventing coronary heart disease

Current research on understanding the causes of coronary heart disease

Current research on treatments for coronary heart disease

Coronary heart disease research labs at the NHLBI

Related coronary heart disease programs

Explore more NHLBI research on coronary heart disease

Cardiovascular disease (CVD) and associated risk factors among older adults in six low-and middle-income countries: results from SAGE Wave 1

Sample and procedure

CVDs conditions

Sociodemographic variables

Risk factors

Alcohol consumption

Physical activity

Fruit and vegetable consumption

Hypertension

Statistical methods

Conclusions

Abbreviations

Acknowledgements

Availability of data and materials

Author information

Authors and Affiliations

Contributions

Corresponding authors

Ethics declarations

Competing interests

Publisher’s Note

Rights and permissions

About this article

Share this article

BMC Public Health

Early prediction of heart disease with data analysis using supervised learning with stochastic gradient boosting

Introduction

Literature survey

Proposed methodology

DTRF classifier

SGB loss optimization

Results and discussion

Performance evaluation

Conclusions

Availability of data and materials

Abbreviations

Acknowledgements

Author information

Contributions

Corresponding author

Ethics declarations

Additional information

Rights and permissions

About this article

Share this article

Signs of heart disease may be more subtle in women than men

Heart attack

Heart failure

Heart valve disease

Abnormal heart rhythm

Peripheral vascular disease

These foods can reduce pain from period cramps

What is cycle syncing? Everything to know, from foods to workouts to fertility

Jennifer Lopez and Ben Affleck split: What’s your divorce risk by years of marriage?

I was battling cancer while starring on Broadway and no one knew

Woman, 106, still lives in her own apartment, eats these healthy foods every day

Common endometrial cancer test is less effective for Black women, research shows

1 exercise alerts woman to 'scary' weakness after extreme reaction to cholesterol drug

Woman shocked when eye freckle turns into melanoma

During labor, her heart stopped for 30 seconds. She had a rare reaction to her amniotic fluid

College student, 20, experienced numbness in hands, feet. It helped doctors find ovarian cancer

Machine learning–based heart disease prediction system for Indian population: An exploratory study done in South India

Bondu Venkateswarlu