Doc Writes Book With ChatGPT; Can AI Detect STIs? How AI Replies to Patient Messages

Welcome to MedAI Roundup, highlighting the latest news and research in healthcare-related artificial intelligence each month.

Robert Pearl, MD, of Stanford University, wrote a book on the potential of AI in healthcare -- and his co-author is ChatGPT. (Axios)

Can an AI company detect sexually transmitted infections (STIs) just from a snapshot of someone's genitals? (STAT News)

Using generative AI to draft physician replies to patient messages resulted in significantly increased read times (likely due to the need to read both the original message and the draft reply), with no change in reply time, according to a study in JAMA Network Open.

Most physicians (68%) said they've changed their views in the last year and now see generative AI as beneficial to healthcare, according to a survey of physicians by Wolters Kluwer.

HCA Healthcare, the nation's largest for-profit health system, is planning to expand the use of ambient clinical documentation technology to more of its emergency departments. (STAT News)

A generative AI clinical documentation tool from Abridge will be deployed at yet another health system; this time at MemorialCare in southern California. Its technology has previously been deployed at Sutter Health, Yale New Haven Health System, UCI Health, Emory Healthcare, the University of Kansas Health System, UPMC, and others, according to a press release.

AI feedback using GPT-4 provided within clinical documentation improved clinical assessments and reasoning across several specialties, including internal medicine, pediatrics, general surgery, and critical care, according to a study in NEJM Catalyst.

A generative AI model beat out ChatGPT at screening for posttraumatic stress disorder among women who recently gave birth, according to a study in Nature.

An AI-based video biomarker was independently associated with the development and/or progression of aortic stenosis in patients without severe disease, according to a study in JAMA Cardiology.

National Cancer Institute researchers developed an AI tool to predict a patient's response to certain cancer drugs, according to a proof-of-concept study in Nature Cancer.

Comment