Adam Rodman

Thomas Buckley, James A. Diao, Pranav Rajpurkar et al.

Multimodal foundation models have shown compelling but conflicting performance in medical image interpretation. However, the mechanisms by which these models integrate and prioritize different data modalities, including images and text, remain poorly understood. Here, using a diverse collection of 1014 multimodal medical cases, we evaluate the unimodal and multimodal image interpretation abilities of proprietary (GPT-4, Gemini Pro 1.0) and open-source (Llama-3.2-90B, LLaVA-Med-v1.5) multimodal foundational models with and without the use of text descriptions. Across all models, image predictions were largely driven by exploiting text, with accuracy increasing monotonically with the amount of informative text. By contrast, human performance on medical image interpretation did not improve with informative text. Exploitation of text is a double-edged sword; we show that even mild suggestions of an incorrect diagnosis in text diminishes image-based classification, reducing performance dramatically in cases the model could previously answer with images alone. Finally, we conducted a physician evaluation of model performance on long-form medical cases, finding that the provision of images either reduced or had no effect on model performance when text is already highly informative. Our results suggest that multimodal AI models may be useful in medical diagnostic reasoning but that their accuracy is largely driven, for better and worse, by their exploitation of text.

Khaled Saab, Jan Freyberg, Chunjong Park et al.

Large Language Models (LLMs) have demonstrated great potential for conducting diagnostic conversations but evaluation has been largely limited to language-only interactions, deviating from the real-world requirements of remote care delivery. Instant messaging platforms permit clinicians and patients to upload and discuss multimodal medical artifacts seamlessly in medical consultation, but the ability of LLMs to reason over such data while preserving other attributes of competent diagnostic conversation remains unknown. Here we advance the conversational diagnosis and management performance of the Articulate Medical Intelligence Explorer (AMIE) through a new capability to gather and interpret multimodal data, and reason about this precisely during consultations. Leveraging Gemini 2.0 Flash, our system implements a state-aware dialogue framework, where conversation flow is dynamically controlled by intermediate model outputs reflecting patient states and evolving diagnoses. Follow-up questions are strategically directed by uncertainty in such patient states, leading to a more structured multimodal history-taking process that emulates experienced clinicians. We compared AMIE to primary care physicians (PCPs) in a randomized, blinded, OSCE-style study of chat-based consultations with patient actors. We constructed 105 evaluation scenarios using artifacts like smartphone skin photos, ECGs, and PDFs of clinical documents across diverse conditions and demographics. Our rubric assessed multimodal capabilities and other clinically meaningful axes like history-taking, diagnostic accuracy, management reasoning, communication, and empathy. Specialist evaluation showed AMIE to be superior to PCPs on 7/9 multimodal and 29/32 non-multimodal axes (including diagnostic accuracy). The results show clear progress in multimodal conversational diagnostic AI, but real-world translation needs further research.

Mohammed Baharoon, Siavash Raissi, John S. Jun et al.

We introduce RadGame, an AI-powered gamified platform for radiology education that targets two core skills: localizing findings and generating reports. Traditional radiology training is based on passive exposure to cases or active practice with real-time input from supervising radiologists, limiting opportunities for immediate and scalable feedback. RadGame addresses this gap by combining gamification with large-scale public datasets and automated, AI-driven feedback that provides clear, structured guidance to human learners. In RadGame Localize, players draw bounding boxes around abnormalities, which are automatically compared to radiologist-drawn annotations from public datasets, and visual explanations are generated by vision-language models for user missed findings. In RadGame Report, players compose findings given a chest X-ray, patient age and indication, and receive structured AI feedback based on radiology report generation metrics, highlighting errors and omissions compared to a radiologist's written ground truth report from public datasets, producing a final performance and style score. In a prospective evaluation, participants using RadGame achieved a 68% improvement in localization accuracy compared to 17% with traditional passive methods and a 31% improvement in report-writing accuracy compared to 4% with traditional methods after seeing the same cases. RadGame highlights the potential of AI-driven gamification to deliver scalable, feedback-rich radiology training and reimagines the application of medical AI resources in education.

David Wu, Fateme Nateghi Haredasht, Saloni Kumar Maharaj et al.

Large language models (LLMs) are routinely used by physicians and patients for medical advice, yet their clinical safety profiles remain poorly characterized. We present NOHARM (Numerous Options Harm Assessment for Risk in Medicine), a benchmark using 100 real primary-care-to-specialist consultation cases to measure harm frequency and severity from LLM-generated medical recommendations. NOHARM covers 10 specialties, with 12,747 expert annotations for 4,249 clinical management options. Across 31 LLMs, severe harm occurs in up to 22.2% (95% CI 21.6-22.8%) of cases, with harms of omission accounting for 76.6% (95% CI 76.4-76.8%) of errors. Safety performance is only moderately correlated (r = 0.61-0.64) with existing AI and medical knowledge benchmarks. The best models outperform generalist physicians on safety (mean difference 9.7%, 95% CI 7.0-12.5%), and a diverse multi-agent approach reduces harm compared to solo models (mean difference 8.0%, 95% CI 4.0-12.1%). Therefore, despite strong performance on existing evaluations, widely used AI models can produce severely harmful medical advice at nontrivial rates, underscoring clinical safety as a distinct performance dimension necessitating explicit measurement.

Ayush Noori, Adam Rodman, Alan Karthikesalingam et al.

Modern computer systems often rely on syslog, a simple, universal protocol that records every critical event across heterogeneous infrastructure. However, healthcare's rapidly growing clinical AI stack has no equivalent. As hospitals rush to pilot large language models and other AI-based clinical decision support tools, we still lack a standard way to record how, when, by whom, and for whom these AI models are used. Without that transparency and visibility, it is challenging to measure real-world performance and outcomes, detect adverse events, or correct bias or dataset drift. In the spirit of syslog, we introduce MedLog, a protocol for event-level logging of clinical AI. Any time an AI model is invoked to interact with a human, interface with another algorithm, or act independently, a MedLog record is created. This record consists of nine core fields: header, model, user, target, inputs, artifacts, outputs, outcomes, and feedback, providing a structured and consistent record of model activity. To encourage early adoption, especially in low-resource settings, and minimize the data footprint, MedLog supports risk-based sampling, lifecycle-aware retention policies, and write-behind caching; detailed traces for complex, agentic, or multi-stage workflows can also be captured under MedLog. MedLog can catalyze the development of new databases and software to store and analyze MedLog records. Realizing this vision would enable continuous surveillance, auditing, and iterative improvement of medical AI, laying the foundation for a new form of digital epidemiology.

Thomas A. Buckley, Riccardo Conci, Peter G. Brodeur et al.

BACKGROUND: For over a century, the New England Journal of Medicine Clinicopathological Conferences (CPCs) have tested the reasoning of expert physicians and, recently, artificial intelligence (AI). However, prior AI evaluations have focused on final diagnoses without addressing the multifaceted reasoning and presentation skills required of expert discussants. METHODS: Using 7102 CPCs (1923-2025) and 1021 Image Challenges (2006-2025), we conducted extensive physician annotation and automated processing to create CPC-Bench, a physician-validated benchmark spanning 10 text-based and multimodal tasks, against which we evaluated leading large language models (LLMs). Then, we developed "Dr. CaBot," an AI discussant designed to produce written and slide-based video presentations using only the case presentation, modeling the role of the human expert in these cases. RESULTS: When challenged with 377 contemporary CPCs, o3 (OpenAI) ranked the final diagnosis first in 60% of cases and within the top ten in 84% of cases, outperforming a 20-physician baseline; next-test selection accuracy reached 98%. Event-level physician annotations quantified AI diagnostic accuracy per unit of information. Performance was lower on literature search and image tasks; o3 and Gemini 2.5 Pro (Google) achieved 67% accuracy on image challenges. In blinded comparisons of CaBot vs. human expert-generated text, physicians misclassified the source of the differential in 46 of 62 (74%) of trials, and scored CaBot more favorably across quality dimensions. To promote research, we are releasing CaBot and CPC-Bench. CONCLUSIONS: LLMs exceed physician performance on complex text-based differential diagnosis and convincingly emulate expert medical presentations, but image interpretation and literature retrieval remain weaker. CPC-Bench and CaBot may enable transparent and continued tracking of progress in medical AI.

Elahe Vedadi, David Barrett, Natalie Harris et al.

Recent work has demonstrated the promise of conversational AI systems for diagnostic dialogue. However, real-world assurance of patient safety means that providing individual diagnoses and treatment plans is considered a regulated activity by licensed professionals. Furthermore, physicians commonly oversee other team members in such activities, including nurse practitioners (NPs) or physician assistants/associates (PAs). Inspired by this, we propose a framework for effective, asynchronous oversight of the Articulate Medical Intelligence Explorer (AMIE) AI system. We propose guardrailed-AMIE (g-AMIE), a multi-agent system that performs history taking within guardrails, abstaining from individualized medical advice. Afterwards, g-AMIE conveys assessments to an overseeing primary care physician (PCP) in a clinician cockpit interface. The PCP provides oversight and retains accountability of the clinical decision. This effectively decouples oversight from intake and can thus happen asynchronously. In a randomized, blinded virtual Objective Structured Clinical Examination (OSCE) of text consultations with asynchronous oversight, we compared g-AMIE to NPs/PAs or a group of PCPs under the same guardrails. Across 60 scenarios, g-AMIE outperformed both groups in performing high-quality intake, summarizing cases, and proposing diagnoses and management plans for the overseeing PCP to review. This resulted in higher quality composite decisions. PCP oversight of g-AMIE was also more time-efficient than standalone PCP consultations in prior work. While our study does not replicate existing clinical practices and likely underestimates clinicians' capabilities, our results demonstrate the promise of asynchronous oversight as a feasible paradigm for diagnostic AI systems to operate under expert human oversight for enhancing real-world care.

Michelle M. Li, Ben Y. Reis, Adam Rodman et al.

Medical AI, including clinical language models, vision-language models, and multimodal health record models, already summarizes notes, answers questions, and supports decisions. Their adaptation to new populations, specialties, or care settings often relies on fine-tuning, prompting, or retrieval from external knowledge bases. These strategies can scale poorly and risk contextual errors: outputs that appear plausible but miss critical patient or situational information. We envision context switching as a solution. Context switching adjusts model reasoning at inference without retraining. Generative models can tailor outputs to patient biology, care setting, or disease. Multimodal models can reason on notes, laboratory results, imaging, and genomics, even when some data are missing or delayed. Agent models can coordinate tools and roles based on tasks and users. In each case, context switching enables medical AI to adapt across specialties, populations, and geographies. It requires advances in data design, model architectures, and evaluation frameworks, and establishes a foundation for medical AI that scales to infinitely many contexts while remaining reliable and suited to real-world care.