Google DeepMind has achieved a major breakthrough in medical AI with its enhanced Articulate Medical Intelligence Explorer (AMIE) system, which can now analyze visual medical data alongside text-based conversations.
The updated AMIE, powered by Google's Gemini 2.0 Flash model, can interpret various medical images including dermatology photos, ECGs, and lab documents during live consultations. This multimodal capability allows the AI to request visual inputs mid-conversation, analyze them, and incorporate findings into its diagnostic process.
In a comprehensive evaluation involving 105 simulated consultations, AMIE demonstrated superior performance compared to board-certified primary care physicians. Medical specialists who reviewed the consultations rated AMIE higher on several critical metrics, including image interpretation quality, diagnostic accuracy, and management plan development.
"AMIE's internal state captures its knowledge about the patient at a given point in the conversation," explained researchers from Google DeepMind in their May 2025 announcement. This enables the system to guide its questioning and responses like an experienced clinician gathering evidence.
Patient actors participating in the study frequently rated interactions with AMIE as superior to those with human doctors, with one noting that AMIE's explanation of a skin condition was clearer than what they received during a recent telehealth visit.
The development comes amid Stanford University's 2025 AI Index report highlighting unprecedented advancement in AI healthcare applications. The report noted that FDA-approved AI-enabled medical devices have skyrocketed from just six in 2015 to 223 by 2023.
While AMIE's capabilities are impressive, Google emphasizes that the system remains experimental. Real-world testing is now beginning through a research partnership with Beth Israel Deaconess Medical Center, where AMIE will be evaluated in actual clinical settings with patient consent.
"The integration of multimodal perception and reasoning marks a helpful step forward for capabilities of conversational AI in medicine," stated Google researchers. "By enabling AMIE to 'see' and interpret the kinds of visual evidence crucial to clinical practice, this research demonstrates the AI capabilities needed to more effectively assist patients and clinicians in high-quality care."