 The study reports on the development and validation of a machine learning model that predicts mortality in COVID-19 patients based on routine laboratory measurements and clinical covariates available within 72 hours of a patient's first positive COVID-19 nucleic acid test result. The GRUD recurrent neural network achieved peak cross-validation performance with an area under the receiver operating characteristic OROC curve of 0.938 and retained strong performance by reducing the follow-up time to 12 hours. The leave one-out feature importance analysis identified age, Charlson Comorbidity Index, minimum oxygen saturation, fibrinogen level, and serum iron level as the most independently valuable features. In the prospective testing cohort, the model provided an OROC of 0.901 and a statistically significant difference in survival, P less than 0.001. The study concludes that the deep learning approach using GRUD provides an alert system to flag mortality for COVID-19 positive patients within a 72-hour window after the first positive nucleic acid test result. This article was authored by Siranya Sankarnarayanan, Jagad Heshwarbalan, Jesse R. Walsh, and others. We are article.tv, links in the description below.