JAMA+ AI Conversations

Arjun K. Manrai, from Harvard Medical School, discusses the intriguing findings of a recent study comparing open-source and proprietary large language models for complex medical diagnoses. He highlights how institutions can utilize custom open-source models while maintaining data privacy. The conversation dives into the competitiveness of models like LLAMA 3.1 against GPT-4, the implications for healthcare technology investment, and the critical role of AI in clinical practice, emphasizing the necessity of human oversight in ensuring diagnostic reliability.

Daniel Herman, MD, PhD, from the University of Pennsylvania, dives into the complexities of prenatal screening for open neural tube defects. He discusses the historical implications of race in these screenings, revealing how racial adjustments can lead to misleading false positives, particularly for Black patients. The conversation highlights a significant shift towards race-agnostic approaches, advocating for more equitable prenatal testing methods. Herman emphasizes the necessity for ongoing research to explore new biomarkers that can improve healthcare outcomes for all.

Artificial intelligence (AI) in health care is advancing, despite concerns about how its use may impact health disparities. Dimitri Christakis, MD, MPH, chief health officer at Special Olympics, joins JAMA Associate Editor Yulin Hswen, ScD, MPH, to discuss AI’s potential role in improving health care delivery for people with intellectual and developmental disabilities. Related Content: How AI Could Improve Health Care for People With Intellectual and Developmental Disabilities How Artificial Intelligence Can Promote Inclusive Health

A recent study showed AI-assisted screening using a large language model tool reduced time to determine trial eligibility compared with manual methods. Author Alexander J. Blood, MD, MSc, cardiologist at Brigham and Women's Hospital, and Associate Director of the Accelerator for Clinical Transformation Research Group at Harvard Medical School joins JAMA Associate Editor Yulin Hswen, ScD, MPH, to discuss this topic and more. Related Content: Study Finds AI Can Quickly Prescreen Patients for Clinical Trials, Speeding Enrollment Manual vs AI-Assisted Prescreening for Trial Eligibility Using Large Language Models—A Randomized Clinical Trial

Join Søren Dinesen Østergaard, a professor at Aarhus University, as he dives into the groundbreaking use of machine learning in predicting the onset of schizophrenia and bipolar disorder. He discusses the significant hurdles in early diagnosis and how timely interventions can improve outcomes. Østergaard also highlights the challenges of varying predictive model performance across hospitals and the need for dynamic, individualized approaches that integrate clinical data for better accuracy. It's a fascinating look at the future of mental health diagnosis!

AI can play a role in addressing language barriers in health care. In a recent Editorial in JAMA Network Open, Pilar Ortega, MD, MGM, of the University of Illinois College of Medicine, and coauthors emphasized the urgent need for integrating language equity into digital health solutions. Dr Ortega joins JAMA and JAMA+ AI Associate Editor Yulin Hswen, ScD, MPH, to discuss. Related Content: Researcher Proposes New Framework for Language Equity in Health Technology Language Equity in Health Technology for Patients With Non–English Language Preference Challenges to Video Visits for Patients With Non–English Language Preference

Cristiana Baloescu, an Assistant Professor of Emergency Medicine at Yale, specializes in using machine learning to improve ultrasound techniques. In this discussion, she unveils how AI can assist non-experts in obtaining diagnostic-quality lung ultrasound images. The conversation dives into AI's role in diagnosing respiratory issues like heart failure and COPD, enhancing timely treatment. Baloescu also outlines the hurdles of integrating AI in clinical settings, emphasizing its potential for improving care in diverse healthcare environments.

A recent study in JAMA Network Open evaluates the use of machine learning algorithms to assess the management of urinary tract infection (UTI). Author Sanjat Kanjilal, MD, MPH, professor in the Department of Population Medicine at Harvard Medical School and Harvard Pilgrim Healthcare Institute, joins JAMA Associate Editor Yulin Hswen, ScD, MPH, to discuss this topic and more. Related Content: Researchers Use Machine Learning to Put Older Clinical Guidelines to the Test Use of Machine Learning to Assess the Management of Uncomplicated Urinary Tract Infection

Digital health technologies, including patient portals, are widely used by older adults, as described in a recent study published in JAMA Network Open. Author Cornelius James, MD, of the University of Michigan joins JAMA+ AI Editor in Chief Roy H. Perlis, MD, MSc, to discuss the study and how it fits with his own experience in the clinic. Related Content: Study Finds Most Older Adults Use Digital Health Technologies, Plus Some Surprises Use of Digital Health Technologies by Older US Adults

Eleni Linos, a prominent dermatologist and epidemiologist at Stanford, shares insights from her recent research on patient satisfaction with AI-generated responses to clinician messages. The conversation reveals that patients often prefer AI for its efficiency over human replies, highlighting a potential shift in healthcare communication. They also discuss how AI can help reduce clinician burnout while balancing patient expectations. Eleni emphasizes the importance of integrating AI responsibly to ensure transparency and compliance in medical interactions.