Using Machine Learning to Predict Risk of Cardiovascular Disease Within 10 Years

Cardiovascular disease (CVD) has emerged as one of the most primary causes of death in South Asia. Therefore, it is extremely important to understand the underlying important factors to prevent disease, as well as acknowledging our current risk in order to potentially treat the disease. Machine Learning is a branch of Artificial Intelligence (AI) which uses computers to analyze and detect patterns in a dataset, using mathematics to create conclusions. This research paper focuses on using Machine Learning to detect CVD risk score, a percentage probability of developing heart-related diseases within 10 years of a person’s life. The dataset used was ethically collected in Bangladesh, a country heavily impacted by the negative effects of CVD. Four machine learning models are used - linear regression, decision tree, random forest, and XGBoost regressor. The metric for comparison is MAE, or mean absolute error, and is a good indicator because it is easy to digest, telling us exactly how much we are “off” by. Linear regression and random forest performed the best, each with around an MAE of 0.88. The study also concluded that the features total cholesterol, diabetes, and BMI were the three key contributors to development of CVD.