Explainable Gait Abnormality Detection Using Dual-Dataset CNN-LSTM Models

2025-09-19

AI Summary

Existing gait abnormality detection models suffer from poor interpretability, limited generalizability, and reliance on single-source data. To address these limitations, we propose a dual-branch CNN-LSTM framework that jointly processes two complementary modalities: GAVD (1D time-series joint kinematics) and OU-MVLP (3D silhouette video sequences). The first branch employs 1D-CNN coupled with LSTM to model joint-level dynamic motion patterns; the second utilizes 3D-CNN with LSTM to capture spatiotemporal dynamics of gait silhouettes. We integrate SHAP for temporal attribution analysis and Grad-CAM for spatial heatmap localization, achieving dual-dimensional (temporal and spatial) interpretability. Evaluated on an independent test set, our model achieves 98.6% accuracy, significantly outperforming baselines in F1-score and recall. This work establishes a high-accuracy, clinically trustworthy, and fully interpretable paradigm for gait assessment and biometric applications.

Technology Category

• Computer Vision: Multi-modal Vision
• Intelligent Robots: Multimodal Perception & Sensor Fusion
• Machine Learning: Multimodal Learning

Application Category

• Semantics and Knowledge: Data modeling to support human-machine intelligence, including LLMs agents, intelligent system behavior, explanations, and user-friendly interactions
• User Modeling, Personalization and Recommendation: Explainable and interpretable methods for personalization
• Responsible Web: Machine-in-the-loop, human agency and autonomy

Abstract

Gait is a key indicator in diagnosing movement disorders, but most models lack interpretability and rely on single datasets. We propose a dual-branch CNN-LSTM framework a 1D branch on joint-based features from GAVD and a 3D branch on silhouettes from OU-MVLP. Interpretability is provided by SHAP (temporal attributions) and Grad-CAM (spatial localization).On held-out sets, the system achieves 98.6% accuracy with strong recall and F1. This approach advances explainable gait analysis across both clinical and biometric domains.

Problem

• Detecting gait abnormalities with explainable AI methods
• Overcoming limitations of non-interpretable single-dataset models
• Providing both temporal and spatial interpretability for diagnoses

Innovation

• Dual-branch CNN-LSTM framework for gait analysis
• SHAP and Grad-CAM provide interpretability for predictions
• Uses joint-based features and silhouettes from two datasets