CAMCA paper “Federated learning for predicting clinical outcomes in patients with COVID-19” has now been published in Nature Medicine!

Last year, during the COVID19 pandemic, CAMCA researchers developed a deep learning model named CO-RISK which takes both chest X-ray images and EHR data to predict the level of supplemental oxygen a patient with COVID-19 symptoms may need 24 and 72 hours after arriving at the emergency department. 

In this new study, dubbed “EXAM” (for EMR CXR AI Model), CAMCA and NVIDIA brought 20 hospitals across five continents together to train the CO-RISK model using privacy-preserving federated learning techniques. The EXAM model shows extraordinary performance for both accuracy and generalizability. We are working on deploying EXAM in the real-world clinical workflow in Mass General Brigham.

The success of the EXAM initiative demonstrates how federated learning can enable the creation of robust AI models in healthcare without breaking data-sharing constraints.

CAMCA will continue our collaboration with NVIDIA and the healthcare startup Rhino Health on more exciting federated learning projects.

Check the NVIDIA blog and video for more information about the EXAM model.

CAMCA Paper Published in Nature Medicine