The Tuning of Machine Learning Models for The Classification of Cardiovascular Disease

Coronary heart disease (CAD) is a cardiovascular disease (CVD) that is one of the main causes of death in the world, and it is very important to diagnose it early in patients who have this disease. In medical diagnosis, computer assistance can be used, especially machine learning models that have high performance. It is done by preprocessing the data and tuning the model to get high performance. Through data preprocessing, machine learning has improved its performance in terms of classification. However, the model has not had the best performance, so it needs to be tuned. Tuning a model using grid search can improve its performance. This research aims to tune the model with grid search on a machine learning model using the Support Vector Machine (SVM), Random Forest (RF), and K-Nearest Neighbor (K-NN) algorithms to classify CAD or normal patients, where the classification is based on symptom features of patients who have CAD or are normal. The heart disease dataset is a public dataset available on Kaggle of 4238 patients consisting of 14 features, including 13 input and 1 output feature. Of the three tested and compared models, the K-NN model produced the highest performance with evaluation results of accuracy = 0.851, IoU = 0.850, precision = 0.853, recall = 0.995, and F1_score = 0.919. It can be concluded that the best model with grid search tuning is K-NN.