Kernel-Based SVM Classifiers for Multi-Disease Forecasting: A Meta-Analysis

Data is very useful in the modern digital age, enormous data sets have been generated in almost every industry. Details concerning individuals' health are often included in healthcare reports. With this level of clinical knowledge, each patient's unique needs can be assessed and can be treated accordingly. The goal of this research was to compare and analyze a few kernel-based Support Vector Machine (SVM) models for use in medical forecasting. Models using the reduced feature sets of the Renal Disorders, Diabetes Mellitus, and Cardiovascular Disease datasets were analyzed using the SVM-Laplace Radial Bias Function (SVM-LR) method. Other machine learning systems, including SVM-Linear (SVM-L), SVM-Polynomial (SVM-P) andSVM-Gaussian Radial Bias Kernel (SVM-G) were also compared to the SVM-LRmodel to determine where there were similarities and where there were differences. Specificity, sensitivity, precision, misclassification rate, and accuracy were only a few of the measures used to evaluate the efficacy of machine learning methods. The results showed that the SVM-LR method performed well than other SVM kernel methods. The testing results demonstrated an accuracy of 98.9% on the Renal Disorders Disease dataset, 98.9% on the Diabetes mellitus dataset, and 90.9% on the Cardiovascular Disease dataset for the SVM-LR method.