A Multi-Layer Perceptron Model with Random Forest Checking Feature Importance for Determining Cardiovascular Disease Risk in Patients

Abstract

Cardiovascular disease is the leading cause of death in first-world countries. It’s identifiable and oftentimes preventable using basic medical data on each patient. This machine learning study attempts to improve easy access to knowledge about the risk of cardiovascular disease to help patients be prepared for possible future options in their life plans. The model used, based on a multi-layer perceptron classifier, would use a publicly available dataset of 303 patients with basic medical information such as age, sex, chest pain, cholesterol levels, and others, as well as already reported risks of cardiovascular disease. Based on its findings it would then determine the patients’ risk of cardiovascular disease using a 1 as high risk and 0 as low risk with an accuracy rate of 86%. This approach can be applied to help improve not only the efficiency inside of a hospital but also provide patients with greater access to vital information.