DeepMind trains Gemini psychiatric consultations with simulated-patient RL
Google DeepMind used simulated patients to conduct multi-turn RL training on Gemini 3.1 Flash-Lite; in 60 paired cases, 76.7% were judged by psychiatrists in a blinded evaluation to be superior to the base model with the same prompt. In 256 adversarial cases, the rate of acute risk not being escalated fell from 30.5% to 12.9%, meaning that roughly 1 in every 8 cases still had a crisis that was not reported. 32 psychiatrists played the role of patients, and another 15 doctors blindly evaluated the conversation records.
Body
New Google DeepMind paper shows that training a model with RL on simulated patients made it a better psychiatric intake interviewer, as judged by psychiatrists.
training on simulated psychiatric intakes cut missed escalations of acute risk from 30.5% to 12.9% in adversarial cases.
In psychiatry, the intake interview is a main diagnostic tool, and key facts often surface only through trust and careful questions. Most conversational medical AI evaluations have left psychiatric cases out.
They trained Gemini 3.1 Flash-Lite with multi-turn RL against simulated patients who reveal key facts only after rapport and targeted questions. Then 32 psychiatrists role-played patients, while 15 others graded the transcripts blind.
The trained agent beat its identically prompted base model in 76.7% of 60 matched case pairs. On 256 adversarial cases, it failed to escalate acute risks 12.9% of the time, down from 30.5%.
If you build health agents, train them on simulated conversations, but keep clinicians supervising, since about 1 in 8 adversarial crises went unescalated.
– arxiv. org/abs/2610.40476
Title: "Training language models in simulation for psychiatric intake"