Multi-class classification and modeling of the hospitalization status of COVID-19 patients

ABSTRACT

In recent times, the unprecedented surge in the Coronavirus disease 2019 (COVID-19) due to the severe acute respiratory syndrome coronavirus 2 (SARS-CoV-2) has led to several attempts at understanding and containing the outbreak of the pandemic as well as to ultimately eradicate it. Steps taken so far include encouraging the wearing of face masks and shields, municipality restrictions such as work-from-home orders, the development of vaccines by health research institutions among others. It is widely believed that the main mode of transmission of the virus is from human to human. In this paper, we present the multi-class classification and modeling of the hospitalization status of COVID-19 patients by using both machine learning and compartmental mathematical models focusing on critical factors like hospital stay-days (SDs) and admission type based on severity of illness. The classification of hospitalization status of COVID-19 patients is necessary in order to know priority cases and give them prompt attention. Two key machine learning algorithms-the decision tree and random forest, are deployed in our analyses. The Levenberg-Marquardt (L-M) algorithm was used for parameter estimation for the mathematical model. From our results, it is easy to identify high risk patients in order to optimize treatment plans that would lower cost of treatments, reduce the chances of others getting infected and assist logistics teams to optimally allocate hospital resources. Hospital administrations can also be supported in deciding the number of staff and visitors per patient per day in a facility. © 2022 IEEE.