Holistic AI-Based Prediction Model for COVID-19 Drug Efficacy in Patients with Comorbidities

The new coronavirus (COVID-19) outbreak had a severe impact on the health of entire communities and the world economy. Despite the high COVID-19 survival rate, there are more severe instances that end in sadism every day. It is anticipated that early identification of COVID-19 at-risk patients and the implementation of preventative interventions will improve patient survival and lower fatality rates. The research involved analyzing 1,023,426 patient samples taken from the Kaggle repository of 60 features. The data were analyzed using classification algorithms, such as Decision Tree Classifier (DTC), Random Forest Classifier (RFC), Gaussian Navies Bayes (GNB), k-Nearest Neighbors (kNN), and Artificial Neural Network (ANN). Initially, the data were pre-processed using several pre-processing techniques. Additionally, the missing literals are filled using kNN Imputer and SMOTE, Nearmiss, and Resample are used to balance out the data. Experiments were performed using twenty-one features of comorbidity diseases along with drugs and fourteen features of comorbidity diseases are identified as significant for predicting the survival versus the deceased COVID-19 patients.The findings indicate that twenty-one features of the imbalanced precision results for k = 5 demonstrate DTC, RFC and GNB achieve a perfect 100% precision, while ANN achieves 94%. Notably, kNN attains a commendable precision of 86%.