Predicting ICU Admissions based on Preconditions and Comorbidity for SARS COV2 Infected Patients

Hospitals across the globe have severe constraints in regard to ICU facilities, beds, and other life support systems. However, in certain situations including natural calamities, epidemics or pandemics, large-scale accidents, and so on, the requirement for ICU beds and resources immediately gets augmented. During such times, there exists an impending need for an optimum apportioning of ICU admissions and resources so that those patients who need critical care are given at the right point of time. The onslaught of COVID-19 pandemic has exuded a high probability of virus transmissions and subsequent complications in patients with co-morbidities and relevant medical issues, resulting in the exploration and investigation of models that could forecast the need for ICU admissions with a higher degree of accuracy. In this research study, a patient’s pre-condition dataset will be used that is categorical in nature. Feature selection and extractions are implemented and the modified descriptors are provided as input to the model, for evaluating them based on the metrics namely F1-score, accuracy, specificity, and sensitivity. The prime objective is to build a predictive algorithm that will predict prior to the necessity of ICU admissions based on the patient’s comorbidity/ precondition specifically for SARS COV2 infection.