A Robust SVM Framework for Heart Disease Detection Utilizing Advanced Feature Selection Techniques

This study proposes a precise model for heart disease prediction through the utilization of support vector machine classifiers. The methodology begins with general data pre-processing the data which encompasses data cleaning which encompasses confirmation of data cleanness by clearing the data of all smoothness), handling of missing values, correcting data inconsistency where necessary, and handling of logical data errors if any. To overcome the class imbalance issue, the SMOTE technique is used here to generate a synthetic sample from the minority class to increase the ratio of the data set. The fact that there is a high variability when designing the model is also addressed whereby the IQR is used to exclude any extreme values to reduce their influence on the potential model. To improve the applicability of the selected features and also deal with the problem of multicollinearity Recursive elimination Feature (RFE) and Correlation Coefficient analysis are being applied. For the SVM algorithm tested on a SMOTE-balanced dataset with selected features, the accuracy, precision, recall, and F1 scores are 95.24%, 89.43%, 93.64%, and 88.79% respectively. This also enhances the reliability of the heart disease model and, above all significant, turns the model into a definitively clinical model that involves retrieval problems that come with computational approaches and the problem of kernels' choice.