A novel machine learning based screening method for high-risk covid-19 patients based on simple blood exams

Abstract

This paper presents a predictive model to potentially identify high-risk COVID-19 infected patients based on easily analyzed circulatory blood markers. These findings can enable effective and efficient care programs for high-risk patients and periodic monitoring for the low-risk ones, thereby easing the hospital flow of patients and can further be utilized for hospital bed utilization assessment. The present machine learning-based SV-LAR model results in a high 87% f1 score, harmonic mean of 91% precision, and 83% recall to classify COVID-19, infected patients, as high-risk patients needing hospitalization.