A lumbar spondylosis auxiliary diagnosis method based on motion information
By collecting multi-joint motion data using distributed inertial sensors and training the model using a spatiotemporal graph convolutional network, the radiation risks and high costs associated with imaging examinations are addressed. This provides a low-cost, non-invasive auxiliary diagnostic method for lumbar spine diseases, improving the model's generalization performance and clinical reliability.
Patent Information
- Authority / Receiving Office
- CN · China
- Patent Type
- Applications(China)
- Current Assignee / Owner
- INST OF COMPUTING TECH CHINESE ACAD OF SCI
- Filing Date
- 2026-04-28
- Publication Date
- 2026-07-17
AI Technical Summary
Current diagnostic methods for lumbar spine diseases rely on imaging examinations, which pose radiation risks and are costly. Furthermore, they cannot dynamically reflect functional status, and the assessment of clinical symptoms lacks objective quantification. Gait analysis based on IMU cannot be directly used for the analysis of lumbar spine diseases.
Multi-joint motion data were collected using distributed inertial sensors to construct a dataset containing lumbar spine disease labels. A lumbar spine disease auxiliary diagnostic model was trained using a spatiotemporal graph convolutional network to capture the spatial dependence and temporal features between joints and to separate causal features for prediction.
It enables low-cost, non-invasive auxiliary diagnosis of lumbar spine diseases, provides objective kinematic evidence, reduces reliance on the experience of domain experts, and improves the model's generalization performance and clinical credibility.
Smart Images
Figure CN122398205A_ABST