DrGait improves clinical gait analysis with clear AI reasoning
DrGait: Biomechanically Grounded Visual Reasoning for Interpretable Clinical Gait Analysis
Computer Vision and Pattern RecognitionArtificial Intelligence
Summary
Analyzing how people walk can help doctors find health problems, but current AI tools often give unclear answers. The authors made DrGait, a system that helps AI make better and more understandable decisions by combining simple measurements of movement with video evidence. Instead of guessing, DrGait checks its ideas against detailed 3D body models to avoid mistakes. This approach produces more accurate diagnoses and clear reports that doctors can trust and review.
What this means in practice
- •For clinical gait analysts: Generate transparent diagnostic reports by integrating video and biomechanical data for better gait disorder detection.
- •For rehabilitation device developers: Incorporate biomechanically grounded AI reasoning into assistive devices to monitor and adjust patient gait.$Commercial implications: Enables advanced gait-monitoring products that provide interpretable diagnostic feedback to therapists and patients.
Authors
Xiangyu Yin, Shiqi Wang, Abrar Alamri, Yasir Aljohani, Weichen Liu, Goeran Fiedler, Wei Gao
Abstract
Current automated gait analysis for clinical applications relies on uninterpretable black-box classifiers. Although Vision-Language Models (VLMs) offer strong reasoning capabilities, applying them directly to gait videos often leads to hallucinations, because they struggle to measure subtle geometric deviations from raw visual contexts. To address this, we introduce DrGait, a training-free agentic framework that shifts the VLM's role from a direct visual reasoner to a clinical planner. DrGait decouples semantic reasoning from geometric perception through a structured Triage-Verification-Synthesis (TVS) workflow. Given an input video and a set of basic spatiotemporal metrics, the DrGait agent first performs a heuristic triage to propose diagnostic hypotheses, which are then verified by autonomously calling deterministic biomechanical tools that operate on reconstructed 3D mesh trajectories, segmented 2D pose tracks, and event-centered video evidence. Finally, a closed-loop mechanism recursively updates the agent's reasoning context based on the feedback. By anchoring VLM's reasoning in verifiable geometric and temporal measurements, DrGait reduces hallucinations, achieving competitive diagnostic accuracy while generating transparent and audit-ready clinical reports.