Machine Learning aided optimal vaccine prediction for COVID-19 based on patient medical history and vaccine side effects

Abstract

Understanding the factors to identify the optimal COVID-19 vaccine is essential for maximizing its impact and is crucial to vaccination success. Some adverse side effects were observed during vaccination due to vaccine selection without analyzing the patient’s medical history. The correlation between patient medical history and the side effects of the vaccine should be addressed. This project proposes a novel approach to predict the optimal COVID-19 vaccine for a person based on their medical history and analyzing the vaccine side effects caused by each disease found in their medical history. Three different machine learning models (SVM, CNN, and KNN) were trained with medical history and side effects after vaccination of 4153 people from various parts of Bangladesh. This system proposes a weighted scoring system for each COVID- 19 vaccine based on its side effects after vaccination, correlating the patient’s medical history. The proposed system achieved accuracies of 54.01%, 48.49%, and 53.77% for SVM, CNN, and KNN, respectively. The proposed system will play an important role in selecting the most effective and suitable COVID-19 vaccine for people with severe health risks.