Predicting Heart Failure with Machine Learning

Testing the effectiveness of different Machine Learning models at predicting heart failure

Introduction

This investigation involved analyzing patient records and diagnoses from Cleveland Clinic to identify symptoms that highly correlated to heart disease fatalities, then evaluating the effectiveness of 3 machine learning models at predicting heart disease.

An effective model accurately predicts whether a patient, given their described symptoms, is likely to experience heart failure. I used the Logistic Regression model, Decision Tree classifier, and K-Nearest Neighbors algorithm to predict whether a patient would experience heart failure.

Results

We used a correlation matrix to identify chest pain, ST depression, number of major blood vessels, and thalassemia as symptoms with the highest correlation to heart disease.

The Logistic Regression model was most effective at predicting heart disease with an accuracy of 88%. Similarly to what we identified through the correlation matrix, the most important features in the logistic regression model were the number of major blood vessels, chest pain, and thalassemia.

Files

• data.csv
• prediction.ipynb

Citations

• Heart failure clinical records. (2020). UCI Machine Learning Repository. https://doi.org/10.24432/C5Z89R.