MedReflect: Teaching Medical LLMs to Self-Improve via Reflective Correction

TL;DR

MedReflect addresses the challenge of autonomous medical reasoning by teaching LLMs a physician-like reflective mode that self-generates hypotheses, self-verifies, and self-corrects in a single pass. It builds a low-cost reflective dataset via RG-based pinpointing and retrospective QA, then fine-tunes a moderate-sized LLM with special tokens to perform reflection during generation. The approach yields significant accuracy gains on medical QA benchmarks, especially for complex tasks, and can approach the performance of larger proprietary models with minimal data and training overhead. This reflection-centric framework reduces reliance on external retrieval or extensive annotated reasoning data, offering a practical path toward robust, cost-efficient medical reasoning by LLMs.

Abstract

Medical problem solving demands expert knowledge and intricate reasoning. Recent studies of large language models (LLMs) attempt to ease this complexity by introducing external knowledge verification through retrieval-augmented generation or by training on reasoning datasets. However, these approaches suffer from drawbacks such as retrieval overhead and high annotation costs, and they heavily rely on substituted external assistants to reach limited performance in medical field. In this paper, we introduce MedReflect, a generalizable framework designed to inspire LLMs with a physician-like reflective thinking mode. MedReflect generates a single-pass reflection chain that includes initial hypothesis generation, self-questioning, self-answering and decision refinement. This self-verified and self-reflective nature releases large language model's latent capability in medical problem-solving without external retrieval or heavy annotation. We demonstrate that MedReflect enables cost-efficient medical dataset construction: with merely 2,000 randomly sampled training examples and a light fine-tuning, this approach achieves notable absolute accuracy improvements across a series of medical benchmarks while cutting annotation requirements. Our results provide evidence that LLMs can learn to solve specialized medical problems via self-reflection and self-improve, reducing reliance on external supervision and extensive task-specific fine-tuning data.

Figures (6)

• Figure 1: Example for Physician Reasoning Proces
• Figure 2: Data Construction Framework
• Figure 3: Examples of Pinpoint Generation
• Figure 4: Index comparison using various data combination
• Figure 5: Performance comparison across different benchmarks