🩺 Medical NLP

🧪 ICML2026 · 2 paper notes

📌 Same area in other venues: 💬 ACL2026 (14) · 🔬 ICLR2026 (5) · 🤖 AAAI2026 (2) · 🧠 NeurIPS2025 (6)

🔥 Top topics: Medical Imaging ×2

ClinTutor-R1: Advancing Scalable and Robust One-to-Many Alignment in Clinical Socratic Education

This paper proposes ClinTutor-R1, the first Vision-Language Agent designed for one-to-many alignment in clinical Socratic teaching. By constructing the 48k ClinTeach dialogue dataset via the multi-agent simulator ClinEdu, and employing explicit Theory of Mind (ToM) reasoning and three-axis rubric reinforcement learning, the model maintains stable teaching quality even as the student count scales to 10, surpassing baseline models by 20% and achieving performance parity with GPT-4o.

MedCase-Structured: A Text-to-FHIR Dataset for Benchmarking Diagnostic Reasoning in Clinically Realistic EHR Settings

The authors propose a "staged LLM + terminology grounding + repair loop" pipeline to convert free-text medical cases into HL7 FHIR R4 standard-compliant patient bundles. Based on this, they construct the MedCase-Structured dataset from MedCaseReasoning, consisting of 1,408 structured synthetic cases (82.5% success rate). Experiments show that the diagnostic accuracy of GPT-5.4 / Gemini-3.1-Pro / Claude-Opus-4.6 consistently drops by 4–23 points on structured FHIR inputs compared to pure text inputs.