Musculoskeletal Injury Recovery Assessment using Gait Analysis with Ground Reaction Force Sensor.

Area of Science:

• Biomedical Engineering
• Rehabilitation Science
• Machine Learning in Healthcare

Background:

• Monitoring musculoskeletal injury recovery is crucial for effective rehabilitation.
• Automated systems are needed to objectively track patient progress during recovery.
• Gait analysis provides valuable insights into lower limb function and recovery.

Purpose of the Study:

• To develop and validate an automated system for assessing musculoskeletal injury recovery.
• To utilize a Long Short-Term Memory Auto-Encoder (LSTM-AE) model for gait pattern analysis.
• To correlate model reconstruction loss with patient recovery status.

Main Methods:

• Utilized the GaitRec dataset, containing ground reaction force (GRF) data from patients with musculoskeletal impairments.
• Trained a LSTM-AE model on healthy gait patterns to establish a baseline.
• Tested the model with GRF signals from injured patients throughout their rehabilitation, analyzing reconstruction losses.

Main Results:

• The LSTM-AE model demonstrated a decreasing reconstruction loss as patients progressed through rehabilitation.
• This trend of reduced reconstruction loss was consistent across both legs and various impairment categories.
• The findings indicate a measurable correlation between model performance and patient recovery.

Conclusions:

• The proposed LSTM-AE model effectively monitors musculoskeletal injury recovery using gait analysis.
• Reconstruction loss serves as a reliable indicator of patient improvement during rehabilitation.
• This automated approach offers a promising tool for objective and continuous recovery assessment.