|
|
|
|
|
One-Year Anniversary
|
October 18, 2025 |
|
|
|
To mark the one-year anniversary of JAMA+ AI, we’ve invited the JAMA Network AI editors to share articles that they found memorable this year.
This week, AI editor for JAMA Romain Pirracchio, MD, MPH, PhD, and AI editor for JAMA Internal Medicine Cary Gross, MD, share their highlights.
|
|
 Romain Pirracchio, MD, MPH, PhD, is a practicing anesthesiologist and critical care physician. He is a Professor in the University of California San Francisco Department of Anesthesia and Perioperative Care and the Ronald D. Miller Distinguished Professor of Anesthesia and Perioperative Care. In his research, he has a focus on clinical AI, causal inference, and perioperative outcomes research.
|
|
|
Dr Pirracchio’s selection: “Optimal Vasopressin Initiation in Septic Shock: The OVISS Reinforcement Learning Study,” published in JAMA March 18, 2025.
This study, on which I was an author, looked at whether a reinforcement learning model could identify when and whether a second agent, such as vasopressin, should be added to critically ill patients already receiving norepinephrine for septic shock.
The model developed and validated in several observational datasets recommended more frequent and earlier use of vasopressin than average care patterns. This was associated with reduced mortality.
This study reports the first evaluation of reinforcement learning to enhance clinical decision-making for the initiation of vasopressin in septic shock. I am proud to be an author on this study and am already working on deploying this model in my hospital.
|
|
|
|
|
 Cary Gross, MD, is a Professor of Medicine and Public Health, and Director of the National Clinician Scholars Program at Yale. His research addresses comparative effectiveness, quality, and health equity, with a focus on cancer prevention and treatment.
|
|
|
Dr Gross’s selection: “Physician- and Large Language Model–Generated Hospital Discharge Summaries,” published in JAMA Internal Medicine on May 5, 2025.
Hospital discharge summaries are vital form of communication between the inpatient and outpatient care teams, ensuring that key treatment plans, test results, and other follow-up care strategies are effectively conveyed. However, they can be laborious to create, and sometimes key items are omitted.
The authors compared discharge summaries created by large language models with those created by clinicians, and found them to be of similar overall quality. This study suggests that with further refinement and appropriate oversight, LLMs can potentially decrease clinicians’ documentation burden without sacrificing quality.
This robust comparison of accuracy, comprehensiveness and quality — including a careful assessment of potential safety signals — will serve as a model for future work comparing LLM with real-world documentation tasks.
|
|
|
|
|
If you enjoy JAMA+ AI, we invite you to forward this email to a colleague and spread the word.
|
|
|
|
|
|
Thank you for subscribing to JAMA Network email alerts. This message was sent to buiduytam1@gmail.com by updates@email.jamanetwork.com.
To update your contact information, change your email preferences, or unsubscribe, click here.
To ensure you always receive JAMA Network emails, add the email address updates@email.jamanetwork.com to your address book.
To unsubscribe by mail, contact:
JAMA Network
AMA Plaza
330 N Wabash Ave
Chicago, IL 60611
Or call (800) 621-8335.
|
      |
|
©2025 American Medical Association. All rights reserved.
|
|
 |
|
Advertisement
|
|
|
