ICU Admission Predictor for COVID-19 Patients

It is very important to detect severity of COVID-19 in earlier stages of the disease. By doing that, hospitals can optimize their workload and give importance to severe cases.

This model is trained with initial clinical variables of COVID-19 patients and it can predict whether the patients will attend intensive care (ICU) or not. It scored 0.93 and 0.90 ROC AUC score on PCR positive and PCR positive/negative patients respectively. The clinical variables used in this model are; age, sex, hypertension, oxygen saturation, respiratory rate, LDH, CRP, ferritin, troponin, lymphocytes, netrophils, serum poassium levels, ALT, glucose, sodium, albumin, procalcitonin and fever.