A COVID-19 Pulmonary Outcome Clinical Prediction Rule Using Epigenetics

PROJECT SUMMARY/ABSTRACT Although most SARS-CoV-2 infected patients develop mild illness, a minority progress to develop severe pulmonary outcomes. The pathogenesis of COVID-19 pneumonia and associated respiratory failure remains poorly understood. Unlike patients with community-acquired pneumonia, who rapidly develop clinical and radiologic evidence of infection, patients with COVID-19 pneumonia have a several-day interval from the start of infective symptoms to hospitalization with radiographically apparent pneumonia. Predicting which patients who initially present with mild symptoms will remain minimally symptomatic versus those who progress to severe pulmonary outcomes is currently impossible. This is a critical knowledge gap because these patients could be targeted with early critical interventions to improve outcomes and preserve limited resources. The objective of this project is to model and validate a clinical prediction rule that incorporates existing, detailed clinical variables and epigenetic markers derived from our electronic medical record data warehouse to develop the COVID-19 severity clinical prediction rule (COPR). The central hypothesis is that, in patients initially presenting with minimal symptoms, the COPR will predict who will remain minimally symptomatic and who will progress to severe pulmonary outcomes. The Specific Aims therefore include: (1) to identify clinical variables and epigenetic markers to predict progression to severe COVID-19 pulmonary outcomes, and (2) to internally validate this clinical prediction rule. This study will facilitate the efficient use of healthcare resources through the identification of infected individuals early in their disease course and prediction of severe pulmonary outcomes during periods of minimal symptoms. Through this project I will learn how to: 1) develop and validate clinical decision rules, and 2) apply `omics to clinical investigation. This combination clinical- epigenetic variable approach could also be beneficial for the prediction of clinical outcomes in other viral infections and may be remodeled, validated, and deployed for the next pandemic.