Related Experiment Video
Updated: Jul 13, 2026
08:24
Sit-to-stand-and-walk from 120% Knee Height: A Novel Approach to Assess Dynamic Postural Control Independent of Lead-limb
Published on: August 30, 2016
10.2K
Cross-speed and Cross-posture Gait Phase Partitioning Based on Multi-modal Data.
Summary
This study developed generalized gait phase partitioning models for patients with gait abnormality. The Random Forest (RF) classifier and electromyography (EMG) data showed the best performance for rehabilitation.
Area of Science:
- Biomedical Engineering
- Rehabilitation Science
- Clinical Biomechanics
Background:
- Gait abnormality is a common motor symptom in neurological disorders like stroke and Parkinson's disease.
- Accurate gait phase recognition is vital for effective rehabilitation, but challenging due to variable patient speeds and postures.
- The need for robust gait phase partitioning models applicable across diverse conditions is critical.
Purpose of the Study:
- To develop generalized models for gait phase partitioning that are effective across different walking speeds and body postures.
- To evaluate the performance of various classifiers and data modalities for cross-condition gait phase partitioning.
- To compare the effectiveness of intra-subject versus inter-subject models for gait phase analysis.
Main Methods:
- Multimodal gait data (including electromyography - EMG) were collected from patients with gait abnormalities at varying speeds and postures.
- Multiple machine learning classifiers, including Random Forest (RF), were trained to construct cross-condition gait phase partitioning models.
- Model performance was systematically evaluated using the F1 score metric under both intra-subject and inter-subject scenarios.
Main Results:
- The Random Forest (RF) classifier achieved the highest performance with a 91.58% F1 score.
- Electromyography (EMG) data demonstrated superior performance among the evaluated modalities, achieving an 89.15% F1 score.
- The inter-subject model demonstrated more favorable outcomes than the intra-subject model for generalized gait phase partitioning.
Conclusions:
- Generalized gait phase partitioning models can be effectively developed using multimodal data, classifiers like RF, and EMG signals.
- These models show promise for application across varied speeds and postures, aiding in the rehabilitation of patients with gait abnormalities.
- The findings support the development of adaptable clinical tools to improve gait rehabilitation outcomes.
