AI Tool Unveils Hidden Burden of Long COVID in Health Records

**Researchers at Mass General Brigham have created an AI-based tool that efficiently combs through electronic health records to help doctors pinpoint cases of long COVID—a condition characterized by enduring symptoms like fatigue, chronic cough, and brain fog following a SARS-CoV-2 infection.** Published in the journal Med, this work aspires to ensure more individuals receive care for this debilitating condition. Their findings suggest a higher prevalence of long COVID than previously recognized, estimating an occurrence rate of 22.8% in contrast to the prior 7%, aligning with national trends. **The AI tool employs a technique called 'precision phenotyping,' sifting through patient records from 14 hospitals and 20 community health centers to link symptoms specifically to COVID.** This innovation could serve as a game-changer for clinicians navigating complex cases with many symptoms. It accounts for a diverse demographic, unlike previous methods that favored those with easier healthcare access. The algorithm excelled in accurately identifying long COVID cases by considering a broader range of patient data, outperforming conventional diagnostic codes by 3% while reducing bias. However, this study acknowledges limitations, such as the potential for incomplete health record data and the potential exclusion of patients whose symptoms may stem from the worsening of pre-existing conditions. Researchers plan to explore the algorithm's efficacy in specific patient groups and share it publicly for broader adoption. They hope this development not only leads to enhanced clinical care but also sheds light on the genetic and biochemical underpinnings of long COVID.