Google's AMIE medical AI demonstrates real-time video consultation in first study

Google Research and Google DeepMind have advanced AMIE, a research medical AI system, to demonstrate expert-level capabilities in real-time clinical video consultations. Built on Gemini and Project Astra using a multi-agent architecture, AMIE now interprets visual and auditory cues, guides virtual physical exams, and performs diagnostic reasoning in real time. This represents a significant expansion from text-based interaction, allowing the system to observe physical signs such as cough, gait, and visible signs of discomfort — the kinds of nonverbal observations that physicians routinely integrate into diagnosis and treatment decisions.

In a randomized study, simulated consultations were conducted with patient actors and evaluated by a group of primary care physicians. Clinical evaluators assessed AMIE across four core competencies: history-taking thoroughness, diagnostic accuracy, management appropriateness, and communication quality. AMIE received favorable assessments across all these domains. Beyond clinical evaluation, patient actors who interacted with the system also preferred the video consultation experience compared to text chat, indicating that the addition of visual communication may enhance patient perception of the interaction.

Despite these encouraging results, Google is careful to underscore that AMIE remains in research and that substantial additional work is required before the system can be deployed responsibly in real clinical settings. The company has not provided details on the timeline, regulatory strategy, or concrete next steps toward clinical availability. The study serves as a proof of concept that AI can perform multimodal clinical assessment in real time, but the path from research demonstration to responsible deployment in actual clinical practice remains open.