Google is developing its medical AI for AMIE research from diagnosis to treatment
Providing a diagnosis is the first step in treating a patient. Once a diagnosis is made, the challenge becomes managing the health condition over time – tracking symptoms over time, analyzing guidelines as they are updated and fine-tuning medications.
Research published today in “Nature” demonstrates the power of the Articulate Medical Intelligence Explorer (AMIE), our state-of-the-art AI research system for medical reasoning and conversation, from one-on-one diagnostic conversations to long-term disease management using drug formulas and clinical guidelines.
Leveraging the long-term context capabilities of Gemini models, AMIE disease management combines an empathetic dialog agent for real-time patient discussions and a deep-thinking management agent that includes hundreds of pages of authoritative clinical information. In this blinded study with patient simulations, specialist physicians compared AMIE with 21 primary care physicians. AMIE matched doctors on overall management thinking and scored high on program accuracy and guideline alignment, suggesting that AI could one day support medical care, giving doctors more time to spend with patients.
Next: We explore how AMIE can work in clinical settings, and present a national study to test AI in real-world virtual care.
