In the JAMA, researchers from Harvard analysed 2.5 years’ worth of OpenEvidence usage by 950,000 clinicians: more than half of the queries relate to the choice of treatment.
A team from Harvard Medical School and Brigham and Women's Hospital, led by Anupam Jena and Vishal Patel, published on October 15 in JAMA what they present as the first large-scale study on how caregivers actually use AI. The field of observation: OpenEvidence, a clinical information platform used by some 950,000 physicians, physician assistants and nurse practitioners in the United States. The number of monthly active users increased from 355 in January 2024 to more than 703,000 in June 2026, for a total of 270 million questions, of which a sample of 38.8 million was analyzed.
Treatment before diagnosis
Clinical decisions clearly dominate: 51.9% of requests concern management and the choice of treatment, compared to 13.9% for diagnosis. Practitioners also often ask the tool about subjects that are far from their specialty. Some disciplines account for much more than their share of the workforce: oncology and haematology account for 3.8% of questions for 1.9% of caregivers, internal medicine 27.3% for 14.5%. Conversely, radiologists, orthopaedists and ophthalmologists use it less than their weight would suggest.
The authors emphasize the main limitation of their work: it describes adoption, not its effects. Whether these consultations improve patients' condition remains to be demonstrated, and according to them, this is the next question to be studied.
Read the full article on the original source
Go to the source →