Prediction of Coronary Heart Disease Using Machine Learning

The Project

Background: Coronary Heart Disease (CHD) is the reduction of blood flow to the heart muscle due to blockage in the coronary arteries and is a leading cause of death in the USA. As such, successful CHD predictions will save lives. Machine Learning has been successfully used to predict weather, the stock market, etc. Hypothesis: Machine Learning approaches can be used to predict 10-year CHD risks in patients. Methods: Using Linear Discriminant Analysis in Python coding, I trained a Machine Learning model with a Framingham heart study dataset of more than 3,658 patients. Next, I predicted 10-year CHD risk in 37 patients. Results: The accuracy of the model turned out to be near 89%. In the test population, the model successfully predicted the 10-year CHD risk in the patients. Conclusion: Successful predictions of CHD using the Machine Learning approach will prevent the disease in patients with a high cardiovascular risk.

View this project on YouTube

Team Comments

CHD kills millions of people in the world. Because CHD takes more than a decade to develop, identifying patients who will suffer from this devastating disease based on the risk factors is imperative to prevent CHD. As such I used Machine Learning to predict this disease.

I had to remove the patients with missing values from the Framingham dataset to avoid errors in Python. Another major problem with this study was to find a testing population with the same parameters. I split the data into two parts - training set (80%) and testing set (20%) to solve the problem.

If I had more time, I would collect data from my locality to predict the risk of cardiovascular diseases in Pittsburgh.