Analysis and Comparison for Innovative Prediction Technique of COVID-19 using Support Vector Machine over Neural Network algorithm with Improved Accuracy

Abstract

Aim: The primary purpose of this study is to improve the accuracy of COVID-19 prediction and evaluation. Materials and Methods: This project is based on data extracted from Kaggle’s website, which is separated into two categories. According to the total sample size estimated by clinical.com, each group comprises 20 samples (N=20) for both the Support Vector Machine (SVM) and Neural Network methods, by keeping 0.05 alpha error-threshold, 95% confidence interval, enrolment ratio at 0:1, and G power at 80%. In MatLab 2021a, this entails training the data and verifying 20 validations ranging from 5 to 24. Results: The SPSS Software and Independent sample T-test are used to contrast the accuracy, sensitivity, and precision rates. The Neural Network has 94.55 percent accuracy (P<0.001), 93.11 percent sensitivity (P<0.001), and 95.31 percent precision (P<0.001), compared to 91.25 percent accuracy (P<0.001), 93.93 percent sensitivity (P<0.001), and 86.11 percent precision (P<0.001) for the SVM. Conclusion: The Neural Network algorithm outperforms the SVM approach in terms of results.