Drivers of mortality in COVID ARDS depend on patient sub-type

In our retrospective study of 2,864 COVID-19 Acute Respiratory Distress Syndrome (ARDS) patients, we used a unique approach combining supervised and unsupervised machine learning to identify distinct phenoclusters. Unlike non-COVID-19 ARDS, our findings show that mortality risk factors don't uniformly apply across all phenoclusters. Some factors increase mortality in some patient phenoclusters while decreasing it in others. This heterogeneity in phenoclusters highlights the challenges in finding effective treatments for all ARDS patients, and points to the need for precision medicine approaches.

READ MORE HERE, in our peer-reviewed publication co-authored with our partners at Northwell Health