Google's AMIE matches physicians in chronic disease management, Nature study finds

AMIE, Google's medical AI system, matched the performance of primary care physicians in managing complex health conditions, according to research published in Nature on June 17, 2026. The conversational AI handled disease management tasks—tracking symptoms, adjusting treatment plans, answering patient questions—at a level comparable to human doctors in controlled evaluations.

The study focused on chronic disease scenarios where patients need ongoing monitoring rather than one-time diagnoses. AMIE's design emphasizes natural dialogue, letting it gather context over multiple exchanges the way a physician would during follow-up visits. That longitudinal capability distinguishes it from earlier medical AI efforts that concentrated on reading scans or flagging abnormal lab values. The research arrives as health systems worldwide face primary care shortages and rising chronic disease loads. Medical AI has historically struggled with the open-ended, multi-visit nature of condition management—most models excel at pattern recognition in imaging or genomics but falter when asked to synthesize patient history, current symptoms, and treatment response over weeks or months.

AMIE was first detailed in a 2024 JAMA preprint that showed the system could match or exceed physician performance in diagnostic conversations. The Nature paper extends that work to longitudinal care, a harder problem that requires the AI to track disease progression and adapt recommendations as patient circumstances change. Google has not disclosed whether AMIE uses a proprietary foundation model or a fine-tuned version of an existing Google architecture, and the Nature paper does not specify model size, training data sources, or inference requirements. The company has not announced clinical deployment timelines.