Performance Analysis of Machine Learning Algorithms for Early Prognosis of Cardiac Vascular Disease

Abstract

Cardiovascular disease, also known as heart disease, is on the rise. It is imperative to anticipate possible illnesses in advance, which is a difficult task that demands precision and efficiency. The main objective of this research paper is to identify patients who are at a higher risk of developing heart disease based on specific medical characteristics. To accomplish this, a heart disease prediction model was created that utilizes a patient's medical history to estimate the probability of a heart condition diagnosis. In this research, heart disease will be predicted using the dataset at hand, which includes 14 key attributes used for analysis. This study evaluates 18 machine learning models on a binary classification task using various performance metrics. The KNeighborsClassifier demonstrated the best performance across all metrics, achieving the highest train and test accuracy, precision, recall, F1-score, and AUC among all models. The ExtraTreesClassifier and GaussianProcessClassifier also performed well, while Gaussian Naïve Bayes, LinearSVC, NuSVC, and LogisticRegressionCV performed the worst. These findings suggest that the KNeighborsClassifier is the most suitable model for the binary classification task in question. This prediction can help clinicians analyze illness risk factors and assess patient scenarios. By focusing more on the condition's risk factors, it can be improved even further.