Google DeepMind has introduced AMIE (Video), a real-time video version of its clinical AI system designed to handle synchronous medical consultations. The company said the system can perceive non-verbal clinical cues, guide patient actors through virtual physical examinations, and reason diagnostically in real time.
The update is described in “Towards expert-level medical AI for real-time video consultations”. Google DeepMind said AMIE (Video) is built on Gemini and Project Astra.
According to the source, the move beyond text-based interfaces matters because written chat systems can lose visual and auditory information that physicians use in diagnosis. The company said patients may also struggle to translate complex physical symptoms into text, which can affect care for people with limited digital or health literacy.
Google DeepMind said AMIE has already shown expert-level performance in text-based diagnostic dialogue and has been effective as a differential diagnosis aid for clinicians. It also said the system has been advanced toward treating and managing disease over time.
The company has also extended AMIE toward specialist-level evaluations in oncology, cardiology and ophthalmology, and toward multimodal diagnostic reasoning over images and clinical documents in simulated settings with patient actors.
Google DeepMind said it is also working on translating these research advances into clinical practice through a framework for physician-centered oversight. That work includes a clinical feasibility study with Beth Israel Deaconess Medical Center and an ongoing nationwide randomized study with Included Health.
In the new study, Google DeepMind said it used 100 scenarios, 300 live consultations and a group of 30 board-certified primary care physicians (PCPs). The company said this is the first demonstration of an AI system exhibiting expert-level performance in real-time clinical video consultations.
Source: research.google.
Companies can share verified announcements through Newz9’s international press release submission page.
