Lightnews — Scholar-powered news

NEJM AI

@ai.nejm.org

1.2K followers 91 following 260 posts

NEJM AI, a new monthly journal from the publisher of @nejm.org, explores the cutting-edge applications of artificial intelligence and machine learning in clinical medicine. Online at ai.nejm.org.

Posts Media Videos Starter Packs

Pinned

NEJM AI @ai.nejm.org · Sep 4

As #AI becomes increasingly more involved in clinical decision-making, it is important to understand whose values are being represented.

In our next event, experts will discuss the complex web of priorities behind AI models in health care.

Sign up: nejm.ai/41ZDYDl

The Hidden Priorities in AI-Driven Care
October 9, 2025
Free Virtual Event
NEJM AI

3 2

NEJM AI @ai.nejm.org · 6h

Original Article by M.V. Heinz et al.: Randomized Trial of a Generative AI Chatbot for Mental Health Treatment nejm.ai/43v2mOY

NEJM AI @ai.nejm.org · 6h

Perspective by Timothy G. Heckman, PhD, John C. Markowitz, MD, and Bernadette Davantes Heckman, PhD: A Generative AI Chatbot for Mental Health Treatment: A Step in the Right Direction? nejm.ai/41N193V

NEJM AI @ai.nejm.org · 6h

Perspective by Ilana Gratch, MS, MPhil, and Todd Essig, PhD: A Letter about “Randomized Trial of a Generative AI Chatbot for Mental Health Treatment” nejm.ai/45JDlyI

NEJM AI @ai.nejm.org · 6h

Heinz and colleagues appreciate the thoughtful responses provided by Heckman et al. and Gratch and Essig regarding their recent publication evaluating the generative AI chatbot, Therabot. Read the full letter: nejm.ai/3UQAyin

#AI #MedSky #MLSky

“We appreciate the opportunity to engage in this correspondence and to provide further context for our study design. We emphasize that ours is an initial trial in this space — not the last.”

Perspective
“Response to Letters about ‘Randomized Trial of a Generative AI Chatbot for Mental Health Treatment’” by M.V. Heinz et al.

NEJM AI @ai.nejm.org · 9h

On NEJM AI Grand Rounds, Dr. Karandeep Singh shares what his team learned from studying a sepsis prediction tool. Hear more from Dr. Singh: nejm.ai/ep34

#AI #MedSky #MLSky

NEJM AI @ai.nejm.org · 1d

Although ML has been shown to outperform surrogate decision-makers in predicting patient preferences for CPR, the authors of a new editorial argue that ML cannot replace surrogates in making decisions for incapacitated patients. nejm.ai/3Vysymr

#AI #MedSky #MLSky

Our greatest concern as clinicians is that the data presented in this article will be interpreted by the media and other readers as evidence that ML can replace surrogates in end-of-life decision-making.

“Machine Learning Cannot Replace Surrogate Decision-Makers in Resuscitation Decisions for Incapacitated Patients” by Robert D. Truog, M.D., M.A., and R. Sean Morrison, M.D.

1 2

NEJM AI @ai.nejm.org · 1d

A new study presents the first proof of concept for a machine learning–based patient preference predictor, which, using data from over 1800 Swiss adults, outperformed human surrogates in predicting cardiopulmonary resuscitation preferences. Learn more: nejm.ai/3IAIYrD

#AI #MedSky #MLSky

Figure 1. Confusion Matrices and Beeswarm Summary Plots.

1 5

NEJM AI @ai.nejm.org · 2d

What we say versus what we do. On NEJM AI Grand Rounds, Dr. Karandeep Singh explains why health #AI needs humility about the gap between theory and practice. Learn more: nejm.ai/ep34

#MedSky #MLSky

NEJM AI @ai.nejm.org · 5d

The authors of a new editorial write that although computational tools cannot address biases inherent in the process of generating observational data, observational emulations may still be valuable for informing trial design and improving trial efficiency. nejm.ai/4mqiyXj

#AI #MedSky #MLSky

“In the presence of unmeasured confounding or other biases that emulation alone cannot address, using emulation estimates as the basis for simulations is imperfect but still preferable to less data-driven approaches for trial design.”

Editorial
“Trial Emulation, Simulation, and Augmentation Using Electronic Health Records and Generative AI” by Issa J. Dahabreh, M.D., Sc.D., Robert W. Yeh, M.D., M.Sc., M.B.A., and Piersilvio De Bartolomeis, M.Sc.

NEJM AI @ai.nejm.org · 5d

Original Article by J. González et al.: TRIALSCOPE — A Framework for Clinical Trial Simulation from Real-World Data nejm.ai/4nnl1TU

NEJM AI @ai.nejm.org · 5d

A new editorial explores whether TRIALSCOPE, or similar applications, could generate sufficient high-quality clinical evidence such that new treatments could be released for widespread use after small, controlled trials. Full editorial: nejm.ai/4nnlrJY

“Oncology is in the unique position of having many therapeutics repurposed and reevaluated for many cancers other than the initial targets. Such a framework is ripe for clinical trial emulation....”

Editorial
“Scoping Out Clinical Trial Emulation” by David Ouyang, M.D., and Jeffrey M. Drazen, M.D.

NEJM AI @ai.nejm.org · 5d

When TRIALSCOPE was applied to over 1 million cancer patient records, the framework structured free text automatically, reduced confounding through causal inference techniques, and enabled the simulation of clinical trials across diverse patient cohorts. nejm.ai/4nnl1TU

#AI #MedSky #MLSky

Original Article | Sep 22, 2025
TRIALSCOPE — A Framework for Clinical Trial Simulation from Real-World Data
J. González and Others

A visual representation of TRIALSCOPE

Reposted by NEJM AI

NEJM.org @nejm.org · 6d

A new article in NEJM Catalyst explores AI’s growing impact on carbon emissions, electricity consumption, and costs for health systems; presents a framework for mitigation of environmental harms; and calls for a collaborative effort among health care and industry leaders. nej.md/3VN8X1Y

Article
“In this single use case, although the absolute emissions presented are not exceedingly high, health AI’s footprint can rapidly accumulate when considering hundreds of different tools for communication, clinical decision support, or predictive modeling.”

“Sustainably Advancing Health AI: A Decision Framework to Mitigate the Energy, Emissions, and Cost of AI Implementation” by Anu Ramachandran et al.

Read this article and more in the October 2025 issue of NEJM Catalyst Innovations in Health Care Delivery at catalyst.nejm.org.

2 6

NEJM AI @ai.nejm.org · 6d

Original Article by M.V. Heinz et al.: Randomized Trial of a Generative AI Chatbot for Mental Health Treatment nejm.ai/43v2mOY

Perspective by M.V. Heinz et al.: Response to Letters about “Randomized Trial of a Generative AI Chatbot for Mental Health Treatment” nejm.ai/3UQAyin

NEJM AI @ai.nejm.org · 6d

In a letter responding to trial results by M.V. Heinz et al., the authors write that although they recognize the potential of digital therapeutics and interactive self-help to expand access to mental health treatment, more robust research is needed. nejm.ai/45JDlyI

#AI #MedSky #MLSky

“Digital therapeutics and interactive self-help tools such as Therabot hold great promise for expanding access to mental health support, and we applaud the creators. But that promise is seriously undermined when preliminary research is presented as a significant clinical advance....”

Perspective
“A Letter about ‘Randomized Trial of a Generative AI Chatbot for Mental Health Treatment’” by Ilana Gratch, M.S., M.Phil. and Todd Essig, Ph.D.

1 1

NEJM AI @ai.nejm.org · 6d

Join the following experts during our next event:

Dr. Noa Dagan, Clalit Innovation, Ben-Gurion University

Dr. Benjamin Glicksberg, Icahn School of Medicine

Dr. Isaac Kohane, Harvard Medical School

Bakul Patel, Google

👉 nejm.ai/41ZDYDl

Free Live Web Event | October 9, 2025
The Hidden Priorities in AI-Driven Care

Isaac Kohane, MD, PhD
Noa Dagan, MD, PhD, MPH
Benjamin Glicksberg, PhD
Bakul Patel

1 1

NEJM AI @ai.nejm.org · 8d

Review Article by L. Altucci et al.: Artificial Intelligence and Network Medicine: Path to Precision Medicine nejm.ai/4mtEamy

#AI #MedSky #MLSky

Figure 1. Biological and Biomedical Challenges Addressed through Integration of Network Medicine and Artificial Intelligence.

Figure 2. The Synergistic Relationship between Network Medicine and Artificial Intelligence.

Figure 3. Analytical Strategies for Networks of Heterogeneous (Biological) Information.

NEJM AI @ai.nejm.org · 9d

Programming is powerful, but domain expertise drives impact. On NEJM AI Grand Rounds, Dr. Karandeep Singh shares the lesson that steered him into medicine. Listen to the full episode: nejm.ai/ep34

#AI #MedSky #MLSky

NEJM AI @ai.nejm.org · 9d

Editorial by Zebadiah Kimmel, MD, MBA: Human Movement Is Poised to Drive Real-World Biomarkers nejm.ai/4mAblVt

#AI #MedSky #MLSky

Page 1 of the editorial "Human Movement Is Poised to Drive Real-World Biomarkers"

"Read full article at ai.nejm.org." sits at the bottom

NEJM AI @ai.nejm.org · 12d

Datasets, Benchmarks, and Protocols by M. Afshar et al.: A Novel Playbook for Pragmatic Trial Operations to Monitor and Evaluate Ambient Artificial Intelligence in Clinical Practice nejm.ai/420zOLm

#AI #MedSky #MLSky

Figure 1. Governance across the Initial Implementation Phase with Learning Health System Integration and Subsequent Design for the Clinical Trial Phase for Operations.

Figure 2. Abridge Ambient Software Integration with Epic Electronic Health Record System.

Figure 3. Difference-in-Differences across the 4-Week Window for Ambient AI Initiation and New Standard Note Implementation.

NEJM AI @ai.nejm.org · 12d

𝗗𝗮𝘁𝗮𝘀𝗲𝘁𝘀, 𝗕𝗲𝗻𝗰𝗵𝗺𝗮𝗿𝗸𝘀, 𝗮𝗻𝗱 𝗣𝗿𝗼𝘁𝗼𝗰𝗼𝗹𝘀
Assessment of Large Language Models in Clinical Reasoning: A Novel Benchmarking Study nejm.ai/3KlD5yX

NEJM AI @ai.nejm.org · 12d

𝗣𝗼𝗹𝗶𝗰𝘆 𝗖𝗼𝗿𝗻𝗲𝗿
The Trump Administration’s Recent Policy Proposals Regarding Artificial Intelligence nejm.ai/46Qwqpj

𝗖𝗮𝘀𝗲 𝗦𝘁𝘂𝗱𝘆
Assessing Retrieval-Augmented Large Language Models for Medical Coding nejm.ai/4mxIQH6

NEJM AI @ai.nejm.org · 12d

𝗢𝗿𝗶𝗴𝗶𝗻𝗮𝗹 𝗔𝗿𝘁𝗶𝗰𝗹𝗲𝘀
Machine Learning–Based Patient Preference Prediction: A Proof of Concept nejm.ai/3IAIYrD

TRIALSCOPE — A Framework for Clinical Trial Simulation from Real-World Data nejm.ai/4nnl1TU

1 1

NEJM AI @ai.nejm.org · 12d

A Response to Dr. Ro’s Letter Regarding the Original Article, “AI Opportunistic Coronary Calcium Screening at Veterans Affairs Hospitals” nejm.ai/46QxtWh

NEJM AI @ai.nejm.org · 12d

𝗣𝗲𝗿𝘀𝗽𝗲𝗰𝘁𝗶𝘃𝗲𝘀
Raging against the Machine: The Human Side of the Story nejm.ai/46W1Fzi

Extending Opportunistic AI Calcium Screening to Noncontrast Head CT for Stroke Risk Stratification nejm.ai/4ntZA3G