Cardio-Vascular Disease Prediction using Machine Learning Techniques

Abstract

The main goal of this study is to use Data Mining Method and Artificial Neural Network to develop a system that can automatically and rapidly predict the risk of coronary heart disease (ANN). The IRT Perundurai Medical College and Hospital's master health checkup data on occupational drivers were used to test this idea (PMCH). Analysis for risk identification is performed in the first stage of the hybrid approach suggested in this study, and level prediction is performed in the second. The sensitivity, specificity, precision, receiver operating curve, area under curve, 10-fold cross validation technique, and the F-measure are used for this investigation. The initial step of the study involves thinking about the most common and changeable dangers. Systolic blood pressure, diastolic blood pressure, and body mass index (BMI) are three biophysical variables, whereas fasting blood sugar, postprandial blood sugar, and triglyceride levels are three blood chemical factors (TG). All of these characteristics have a predetermined margin value that is based on WHO guidelines. Support Vector Machine (SVM), Naive Bayes (NB), and the C4.5 algorithm in Decision Tree are the three approaches used to categorize these variables and forecast the risk (DT). The C4.5 algorithm fared best in forecasting CHD risk when the three approaches were compared using the performance metrics, as discovered by the investigation. The decision tree C4.5 method outperformed the other two classifiers with an improved 99.5% accuracy and 99.67% sensitivity. The increased percentage demonstrates that the Decision tree method delivered consistent results that were better to those produced by the Naive Bayes and SVM models.