Artificial intelligence: Medical AI models for patient management (Nature)
18 June 2026
Two independent AI models that can assist with multiple stages of patient management, from diagnosis to treatment decisions, are presented in Nature this week. The systems — MIRA (Medical Intelligence for Reasoning and Action) and Google's AMIE (Articulate Medical Intelligence Explorer) — perform at least as well as physicians, demonstrating the potential for conversational AI tools to help with disease management.
Large language models (LLMs) have shown promising developments for clinical applications, but they tend to specialize in narrowly defined tasks. The clinical management of patients requires a multifaceted approach, delving into patient histories, carrying out appropriate investigations, making accurate diagnoses, planning treatment options (both pharmaceutical and surgical), and monitoring outcomes over multiple visits. If AI agents could carry out such tasks, achieving effective management reasoning, they may be able to assist physicians in routine tasks and possibly address physician shortages in some regions of the world. Two papers in Nature report advances in the capabilities of autonomous medical AI agents.
Jakob Kather and colleagues describe MIRA, an AI model that has access to patient data in an isolated electronic health record system. The model is evaluated using real-world data from more than 500 emergency department clinical cases. MIRA gathers information via chat with a patient AI agent whose responses match documented histories taken from clinical notes. MIRA can choose from over 85,000 options to order diagnostic tests, interpret the results, and make treatment plans including prescribing medication, scheduling procedures and arranging admissions. It achieved an average diagnostic accuracy of 87.8%, compared to 78.1% from a panel of six physicians across specialities. Future work is needed to further improve accuracy and establish generalization in real-world studies, the authors conclude.
Mike Schaekermann and colleagues describe AMIE, an LLM-based system optimized for clinical management and conversations. The model can perform continuous reasoning over multiple patient visits to map the progression of disease and responses to treatment. AMIE uses Gemini to analyse the information retrieved from the patient and align its output with relevant and up-to-date clinical practice guidelines and drug formularies (lists of approved, clinically preferred medications). In a virtual clinical examination study, AMIE was compared to 21 primary care physicians across 100 multi-visit case scenarios and five medical specialities, designed to reflect UK NICE guidance and BMJ Best Practice guidelines. AMIE performed as well as real physicians in management reasoning capabilities, and better than physicians in preciseness of treatments and investigations and in its alignment with clinical guidelines and grounding of management plans in those guidelines. On a newly introduced benchmark for medication reasoning (RxQA), AMIE outperformed physicians on difficult cases. The authors note that more work is needed before AMIE is ready for clinical care but conclude that this work represents a step towards the use of conversational AI tools to assist physicians in disease management.
- Article
- Open access
- Published: 17 June 2026
Ferber, D., Hilgers, L., Höper, C. et al. Towards autonomous medical artificial intelligence agents. Nature (2026). https://doi.org/10.1038/s41586-026-10675-5
- Article
- Published: 17 June 2026
Liévin, V., Palepu, A., Weng, WH. et al. Towards Conversational AI for Disease Management. Nature (2026). https://doi.org/10.1038/s41586-026-10764-5
© 2026 Springer Nature Limited. All Rights Reserved.
