Can Machine Learning Predict Heart Attacks?
The "Data Doctors," a student team from the University of Alabama, put this question to the test.
Competing in the 2025 SAS Curiosity Cup, they analyzed a dataset containing over 429,000 records to identify key risk factors for heart attacks - and their project secured a spot among the top 15 teams in this global competition. Congratulations!
Cardiovascular disease (CVD) is the leading cause of death worldwide, affecting more than 120 million adults in the U.S. and accounting for nearly one-third of all deaths. Heart attacks, a severe form of CVD, strike approximately 800,000 Americans each year.
The Team wanted to develop a predictive model for assessing the likelihood of a heart attack based on lifestyle habits and medical history.
What did they do?
- Used data from Kaggle, originally sourced from the Behavioral Risk Factor Surveillance System - an annual phone survey collecting responses from all U.S. states plus Guam, Puerto Rico, and the Virgin Islands.
- Cleaned and engineered features such as age groups, chronic disease scores, and sleep patterns.
- Tested several models, including Gradient Boosting, Bayesian Networks, Neural Networks, and more, using SAS Viya for Learners.
- Achieved their best results (88.4% accuracy) with an Ensemble model combining Gradient Boosting, Bayesian Network, and Generalized Additive Model techniques.
- Identified key predictors of heart attack risk: Age Group, Cardiac Risk Score, Chronic Disease Score, Diabetes, Kidney Disease, and Angina.
Their findings demonstrate that AI-powered models can play a crucial role in the early identification of high-risk patients, provided that we balance model performance with interpretability and trust.
The Data Doctors team: Alec Scheffler, Tre Turner, and their Faculty advisor: Yana Melnykov. Their full paper is available as a PDF below the headline.
If you would like to participate in the new edition of The Curiosity Cup competition, please note that registration will be open from October 2025. Follow the website https://www.sas.com/en_us/events/23q3/curiosity-cup.html
