Clinical Decision-Making

Large language models (LLMs) show great potential for clinical decision-making, yet most applications remain narrow, task-specific chat tools rather than systems integrated into clinical workflows1,2. However, building physician copilots will require models that operate within the electronic health record (EHR), with governed access to patient data and the ability to initiate permitted EHR actions within defined safety constraints. Yet it remains unproven whether such a system can manage patient cases with physician-level performance. Here we show that MIRA (Medical Intelligence for Reasoning and Action), an autonomous artificial intelligence agent operating in a sandboxed EHR environment, can navigate a large clinical action space to obtain patient histories; order and interpret laboratory, imaging and microbiology tests; generate differential diagnoses; and formulate treatment plans such as prescribing medications, scheduling surgical procedures and planning admissions. In simulations on real patient cases spanning multiple diagnoses, MIRA outperformed physicians in diagnostic accuracy and made guideline-concordant, medication-safe and appropriate admission decisions. Compared with previous LLM applications that addressed isolated subtasks or provided free-text advice, these results suggest that an EHR-integrated artificial intelligence agent can turn clinical intent into structured, actionable EHR operations, possibly making it a more effective decision-support partner for physicians. Further work is needed to establish generalization, safety and governance through prospective, real-world studies.

ImportanceLarge language models (LLMs) are increasingly marketed for clinical use, yet their ability to replicate full-spectrum clinical reasoning remains uncertain. Existing evaluations often rely on multiple-choice examinations that do not reflect the complexity of patient care.ObjectivesTo evaluate the longitudinal clinical reasoning ability of state-of-the-art LLMs and to introduce a multidimensional, clinically meaningful benchmark for clinical-grade artificial intelligence (AI).Design, setting, and participantsIn this cross-sectional study, performance was evaluated using standardized clinical vignettes from the January 2025 update of MSD Manual vignettes. A total of 21 off-the-shelf LLMs, including recently released GPT-5, Claude 4.5 Opus, Gemini 3.0 Flash and Pro, and Grok 4, were evaluated. Models were assessed by medical student scorers in triplicate across sequential stages of the standard clinical workflow. Analyses were performed from January to December 2025.Main outcomes and measuresThe primary outcome was the Proportional Index of Medical Evaluation for LLMs (PrIME-LLM) score, defined as the normalized polygonal area representing balanced accuracy across 5 domains of clinical reasoning as follows: differential diagnosis, diagnostic testing, final diagnosis, management, and miscellaneous clinical reasoning questions. Analyses including analyses of variance, t tests, and regression models were used to compare AI model performance and demographic associations.ResultsLLMs were tested across 29 clinical vignettes (representing 16 254 responses in total). PrIME-LLM scores ranged from 0.64 (range, 0.63-0.65) (Gemini 1.5 Flash) to 0.78 (range, 0.77-0.79) (Grok 4), with reasoning-optimized models outperforming nonreasoning models and GPT models scoring highest overall. Differential diagnosis was less accurate than diagnostic testing, while final diagnosis, management, and miscellaneous reasoning were more accurate. Failure rates exceeded 0.80 (range, 0.90-1.00) for differential diagnosis in all models but were less than 0.40 (range, 0.09-0.39) for final diagnosis. Multimodal performance was robust; most LLM models showed improved accuracy with image inputs.Conclusions and relevanceIn this cross-sectional study of 21 LLMs, frontier LLMs achieved high accuracy on final diagnoses but performed poorly in generating differential diagnoses and navigating uncertainty relative to other reasoning stages. The PrIME-LLM framework provided greater separation than raw accuracy, revealing critical reasoning gaps obscured by traditional benchmarks. Thus, despite version-based improvements and advantages in reasoning-optimized models, off-the-shelf LLMs have not yet achieved the intelligence required for safe deployment and remain limited in demonstrating advanced clinical reasoning.

ImportanceUnderstanding of intraocular pressure (IOP) as a continuous risk factor for glaucoma has evolved over time, and IOP reduction is widely acknowledged as the mainstay of treatment. However, the impact of specific IOP levels on clinical decision-making remains an underexplored topic.ObjectiveTo assess how IOP levels influence the decision to initiate or escalate glaucoma therapy in clinical practice.Design, setting, and participantsIn this retrospective, multicenter cohort study, the Sight Outcomes Research Collaborative (SOURCE) ophthalmology data repository was used to identify clinic encounters between October 2009 and January 2022 for patients with glaucoma with IOPs ranging from 12 mm Hg to 25 mm Hg. Data analysis was performed from July 2024 to September 2025.Main outcomes and measuresThe primary outcome was whether IOP-lowering therapy was initiated or escalated after each clinic encounter, defined as a new prescription for IOP-lowering medication within 1 week of the encounter, laser treatment within 4 weeks, or glaucoma surgery within 8 weeks. The rate of treatment initiation at different IOP levels was measured, and then mixed-effects logistic regression was used to model the odds of treatment initiation at specific indicator IOP levels.ResultsThis analysis included 1 866 801 clinic encounters from 184 504 eyes of 94 232 unique patients across 7 sites in SOURCE. Mean (SD) patient age was 69.5 (10.8) years, and of the total clinic encounters, 1 084 827 (58.1%) included female patients. The rate of IOP-lowering treatment increased with higher IOP levels, with the largest acceleration in treatment rate at IOPs of 22 mm Hg or higher. With mixed-effects logistic regression modeling, an indicator IOP of 22 mm Hg had a greater effect on treatment initiation (odds ratio, 1.11; 95% CI, 1.08-1.14) compared with lower indicator IOPs.Conclusions and relevanceIn this cohort study, while clinicians seem to generally use IOP as a continuous risk factor in their treatment patterns, with higher rates of glaucoma therapy at increasing IOP levels, these findings suggest that the historical IOP cutoff of 22 mm Hg may still influence clinician decision-making in glaucoma management. Improved clinical decision support may be useful to assist clinicians with using IOP as a continuous risk factor in their decision-making.