A Stanford pilot study found that an AI workflow used to draft hospital discharge summaries was associated with low reported harm risk. Physicians incorporated AI-generated text into 57% of discharge notes, and 88% of reviewed summaries were rated as having no harm potential. Omissions were more common than hallucinations, occurring in 25% of reviewed summaries compared with 2% for hallucinations. Physicians reported perceived efficiency gains and lower burnout scores, although objective time savings were modest. Researchers said future systems may need structured electronic health record data to reduce omissions and improve reliability.

Source: JAMA Network Open