“Physicians are often faced with having to wade through a tangled web of symptoms and medical histories, unsure of which threads to pull, while balancing busy caseloads. Having a tool powered by AI that can methodically do it for them could be a game-changer,” said co-lead author of the study, Alaleh Azhir, an internal medicine resident at Brigham and Women’s Hospital, Boston, in a press release.
For the study, researchers analyzed electronic health records from over 295,000 patients from 14 hospitals and 20 community health centers in Massachusetts. The algorithm used an attention mechanism to exclude sequelae that may be explained by prior conditions. The researchers defined long COVID as a diagnosis that could not be explained by a patient’s medical record yet was associated with a COVID-19 infection. Diagnosis needed to have persisted for at least 2 months.
Ultimately, the algorithm had a 79.9% precision compared to that of the official diagnostic code for long COVID standing at 77.8%. The algorithm estimated a long COVID prevalence of 22.8% of the population, which is close to national regional estimates.
“We also provide an in-depth analysis outlining the clinical attributes, encompassing identified lingering effects by organ, comorbidity profiles, and temporal differences in the risk of post-acute sequelae of COVID-19 (PASC). The PASC phenotyping method presented in this study boasts superior precision, accurately gauges the prevalence of PASC without underestimating it, and exhibits less bias in pinpointing long COVID patients,” wrote the researchers in their paper.
The research additionally identified a long COVID research cohort of over 24,000 patients, compared to around 6,000 found when using the official diagnostic code. The researchers noted that the cohort identified via the algorithm will aid research into long COVID’s genetic, metabolomic, and clinical intricacies, overcoming limitations from previous cohort studies, which were impaired by limited size and outcome data.
Sources: Neuroscience News, MedRxiv
