Predicting Cardiovascular Disease with Machine Learning: A Comparative Analysis of Classification Models

Abstract Cardiovascular disease is a very serious health issue. So, in order to prevent its spread, we need to understand the reason behind it’s increase. Different factors like our lifestyle, our genes, the surroundings that we live in and so on all contribute to the risk of getting CVD. So, it is important to make positive changes in our day-to-day life which in the end will make us healthier. This research paper delves into understanding the importance behind such factors. We have used various classification models such as Logistic Regression, Decision Tree Algorithm, Random Forest, KNN, Support vector machine and Naïve Bayes to make predictions regarding cardiovascular disease patients. We have used data from UCI Repository that includes the features (predicator variables) such as age, BMI, gender, cholesterol, alcohol intake and so on to determine the presence of cardiovascular disease patients (response variable). Different models have been used to find out which model works best and we have done this by estimating various metrics that are essential for the assessment of model performance such as accuracy, precision, recall, etc. The Support Vector Machine model had the highest accuracy, Roc-Auc. So, this shows that the Support Vector Machine (SVM) so far is the best model for making predictions regarding CVD.