AI doctor agent uses reinforcement learning for proactive medical consultations

Researchers have developed DoctorAgent-RL, a novel multi-agent reinforcement learning framework designed to improve AI's capabilities in real-world clinical consultations. This system trains a doctor agent, utilizing the Qwen2.5-7B-Instruct model, to proactively gather patient information through strategic questioning rather than relying on static, single-turn interactions. Evaluations, including blinded human assessments and trials with actual patients, demonstrated that DoctorAgent-RL achieved a 70% exact diagnostic match rate, outperforming existing frontier models. AI

IMPACT This AI agent could alleviate physician shortages and reduce misdiagnosis risks by handling initial patient screenings.

RANK_REASON This is a research paper detailing a new framework and dataset for an AI agent in a specific domain.

Read on arXiv cs.CL →

• arXiv
• DoctorAgent-RL
• LLMs
• MTMedDialog
• Qwen2.5-7B-Instruct

• paper
• model release
• product

AI-generated summary · Google Gemini · from 1 sources. How we write summaries →