Jenna Wiens - Model to Estimate Patient Risk

In response to the strain on the hospital system induced by a growing number of COVID-19 cases, our interdisciplinary team composed of computer scientists and clinicians developed M-CURES - the Michigan COVID-19 Utilization and Risk Evaluation System. This machine learning based approach to clinical decision support aims to identify who among patients diagnosed with COVID-19 is at highest (and lowest) risk of requiring intensive care during hospitalization. Applied to a held-out cohort of COVID-19 patients at Michigan Medicine, the proposed model outperformed a widely implemented proprietary deterioration index. We are now exploring the integration of this model into the EHR and clinical workflow. Going forward, such a model could help clinicians during a future surge in hospitalizations or as an early warning system applied to the general inpatient population.