A clinical knowledge-guided attention framework for gait-based adult spinal deformity diagnosis

This research propose a framework that leverages clinical knowledge to guide gait classification for the automated diagnosis of Adult Spinal Deformity (ASD) using monocular gait videos. Existing models often overlook anatomically meaningful features, limiting their clinical applicability. This research introduce clinician-informed attention maps to encode expert knowledge about diagnostic joints and motion patterns. These maps guide the spatiotemporal focus of a CNN-based backbone to attend to clinically relevant cues. Experiments show that our approach improves both accuracy and interpretability over traditional baselines, demonstrating its potential as a non-invasive and explainable screening tool for ASD.