Infection-Reasoner: A Compact Vision-Language Model for Wound Infection Classification with Evidence-Grounded Clinical Reasoning

TL;DR

Infection-Reasoner is a compact vision-language model that classifies wound infections and provides evidence-grounded explanations, trained via a two-stage process involving GPT-5.1 and reinforcement learning, achieving high accuracy and interpretability.

Contribution

The paper introduces Infection-Reasoner, a novel reasoning vision-language model with a two-stage training pipeline for wound infection classification and explanation generation.

Findings

Achieved 86.8% accuracy on wound dataset

Rationales rated 61.8% as Correct by experts

Visual-support agreement scores ranged from 0.722 to 0.903

Abstract

Assessing chronic wound infection from photographs is challenging because visual appearance varies across wound etiologies, anatomical locations, and imaging conditions. Prior image-based deep learning methods have mainly focused on classification with limited interpretability, despite the need for evidence-grounded explanations to support point-of-care decision making. We present Infection-Reasoner, a compact 4B-parameter reasoning vision-language model for chronic wound infection classification and rationale generation. To address the scarcity of expert-labeled wound images with reasoning annotations, Infection-Reasoner is trained using a two-stage pipeline: (1) reasoning distillation, in which GPT-5.1 generates chain-of-thought rationales for unlabeled wound images to initialize wound-specific reasoning in a smaller student model (Qwen3-VL-4B-Thinking), and (2) reinforcement learning post-training with Group Relative Policy Optimization on a small labeled infection dataset to refine classification reasoning. On a held-out heterogeneous wound dataset, Infection-Reasoner achieved 86.8\% accuracy, 86.4\% sensitivity, and 87.1\% specificity, outperforming several strong baselines, including GPT-5.1. Rationale quality was further evaluated using both multimodal large language model (MLLM) judges and wound expert review. Across four MLLM judges, visual-support agreement scores ranged from 0.722 to 0.903, while expert review rated 61.8\% of rationales as Correct and 32.4\% as Partially Correct.