Stony Brook researchers found doctors preferred an AI-written hospital discharge summary in 95% of cases reviewed, though a doctor must still check every draft.
Doctors reviewing hospital discharge summaries picked the AI-written version 95% of the time in a new Stony Brook University study. The tool drafted the notes that sum up a patient's hospital stay. The work was published in Digital Medicine, a Nature partner journal. The researchers say a physician must still review any AI draft before it goes into a patient's record.
A discharge summary is handed off when you leave the hospital. It covers treatment, medication needs, changes in medication, unresolved medical issues and what follow-up care is needed as an outpatient. Primary care doctors rely on it to continue your care. It works as a roadmap for what comes next. Long, complex stays produce a lot of information, and condensing it takes time and effort.
Researchers compared AI-written and doctor-written summaries from 60 internal medicine stays at Stony Brook University Hospital. Each stay lasted seven to 21 days. Hospital-based doctors and primary care doctors evaluated both versions. The AI was a large language model, a type of AI that reads and writes text. It was developed with input from clinicians and turned medical records into concise summaries within minutes.
Both kinds of summaries contained errors. Reviewers found no statistically significant difference between them in how severe the errors might be or how likely they were to cause harm. That means the AI was not error-free. The authors stress that physician review and verification remain essential before any AI-generated discharge draft is used in a patient record.
Dr. Rachel Wong, a co-corresponding author, said the results provide evidence that AI can support discharge documentation for complex hospital stays.
The tool is not yet part of everyday care. Co-corresponding author Fusheng Wang said the team will keep developing it for complex discharge summaries. Next steps include checking accuracy, safety, how much time it really saves physicians and how it would fit into daily hospital work. A pilot study in everyday clinical practice at Stony Brook Medicine is being considered. Whether it saves doctors time in practice is not yet known.
AI drafted discharge notes that doctors preferred, but errors still appeared and a doctor must check each one before it reaches your record.
World models, AI that learns physics from video, are being built into home robots and driverless cars, and your own videos may help train them.
Read more: AI that learns how the physical world works is coming to robots and carsA new Alibaba-led test of real coding requests found the best AI agent fully finished just 26.4 percent of tasks, so human review still matters.
Read more: AI coding helpers fully finish only about 1 in 4 real tasksA University of Chicago law professor asked Claude to fix page numbers. It deleted his syllabus and a year of notes instead. Dropbox saved him, not the AI.
Read more: An AI assistant deleted a law professor's year of lecture notes
