Comparative Analysis of Kernel based Support Vector Machine Models for Multi Disease Prediction

Abstract

In today's digital age, data is invaluable, and massive amounts of data have been produced in every field imaginable. Reports from the healthcare sector often contain details about patients' health. By having this clinical expertise, we are better able to detect undetectable health problems and provide individualised therapy to each patient. The purpose of this study was to evaluate and contrast several kernel-based Support Vector Machine (SVM) models for use in healthcare prognostication. With the SVM-LRBF technique, we examined the models with the feature reduction set of the Renal Disorders Disease, Diabetes Mellitus, and Cardiovascular Disease datasets. Similarities and differences between the models and other machine learning systems such as Random Forest, SVM-Linear, Decision Tree, SVM-Gaussian Radial Bias Kernel, and SVM-Polynomial were also analysed. Performance of machine learning approaches was measured using a number of different metrics, including specificity, sensitivity, precision, misclassification rate, and accuracy. The experimental findings showed 98.1 percent accuracy for the Renal Disorders Disease dataset, 90.9 percent accuracy for the Diabetes mellitus dataset, and 98.1 percent accuracy for the Cardiovascular Disease dataset.