Patient AI in Primary Care: 114 Patients, 0 Stops [2026]
A study by BIDMC and Google tested AI chatbot AMIE in primary care, with 114 patients and no safety stops. AMIE's diagnostic accuracy was 90% in top 7, 75% in top 3, and 56% for the top diagnosis. Alphabet funded the study, which ran from April to November 2025.
How this was made
The 30-second read
Why it matters
The study provides early safety validation for AI chatbots in healthcare, potentially influencing investor sentiment toward Alphabet's health AI initiatives.
Market read
While the trial is a positive signal for health‑AI, its early stage and limited scope suggest only modest market impact.
What to watch
Lack of control group, small patient cohort, and model changes mid‑trial limit the robustness of the results.
Background
The article reports on a first‑of‑its‑kind clinical feasibility study of Google's conversational AI AMIE in urgent care, funded by Alphabet and conducted at a Boston medical center.
Ticker impact
Alphabet-funded study shows its AI chatbot AMIE had zero safety stops in 98 supervised patient encounters, indicating safety in primary care.
modest upside as investors view safety validation favorably, but limited by early-stage trial scope
First disclosure of clinical safety data; however, no control group and limited scale temper market reaction.
Market effects
May encourage further investment in AI-driven healthcare tools across the med‑tech sector.
Limited to US tech and health‑care investors; no immediate regional shift.
Sets a precedent for AI health applications worldwide, but impact remains modest.
Counterpoint
Safety in a supervised setting does not guarantee performance in real‑world unsupervised use; investors should stay cautious.
Key entities
- CompanyAlphabet
Parent company funding the AI health study.
- CompanyGoogle
Developer of the AMIE conversational AI.
- InstitutionBeth Israel Deaconess Medical Center
Site where the trial was conducted.
