Google tested its AMIE medical chatbot with 100 urgent-care patients

In short: Google ran an early test where 100 urgent-care patients chatted by text with its AMIE medical chatbot before seeing a clinician.

What happened

AMIE, short for Articulate Medical Intelligence Explorer, is a research chatbot from Google built to talk with patients about symptoms and suggest possible causes. Think of it like a structured intake form that can ask follow-up questions, instead of a paper checklist.

In the study, 100 adults who were already scheduled for urgent-care visits had a text conversation with AMIE before their appointment, in some cases up to five days earlier. The chatbot asked about symptoms and medical history, then produced a list of possible diagnoses for the patient to discuss with their clinician.

The conversations were monitored by human safety supervisors. The goal was to see if this kind of workflow is practical and safe, not to replace clinicians or let the chatbot make final medical decisions.

Why it matters

In a later review, evaluators compared AMIE’s suggestions with clinicians’ work and what ended up in patients’ medical records. They reported no statistically significant difference, which means the study did not find a clear, measurable gap between AMIE and clinicians on overall quality of the list of possible diagnoses. The reviewers also found no clear differences in how appropriate or safe the suggested care plans were.

This is still a small, early study in one urgent-care setup. It does not prove AMIE is equal to a doctor, and it does not show better health outcomes or less work for staff. But it does show how AI might help collect information before a visit, so the appointment can focus on decisions instead of basic questions.

Source: NYTimes