Jay Lee - October 7, 2020

Jay Lee, PhD, Staff Associate, Laboratory of Dr. Chunhua Weng, Department of Biomedical Informatics & Cong Liu, PhD, Associate Research Scientist, Laboratory of Dr. Chunhua Weng, Department of Biomedical Informatics, Columbia University, "Severity Prediction for COVID-19 Patients via Recurrent Neural Networks"

A puzzling property of COVID-19 is its separation into asymptomatic and very mild cases and very severe cases, up to death. While certain risk factors for a severe progression were identified, no reliable prediction, especially at an early stage of the disease can be made. The predictive model developed by Dr. Lee and Dr. Liu is trying to do this. Focussing on EHR data as input the model, derived from 2500 hospitalised patients of 18 years and older needs only a fraction of a second per patient to predict the severity of the disease, which makes it usable for real time applications. Already now it is more reliable than for example age, the biggest risk factor for COVID-19.