Related Experiment Video
Updated: Dec 16, 2025
06:28
Biomechanical Changes Related to Low Back Pain: An Innovative Tool for Movement Pattern Assessment and Treatment Evaluation in Rehabilitation
Published on: December 13, 2024
970
Using a Motion Sensor to Categorize Nonspecific Low Back Pain Patients: A Machine Learning Approach
Masoud Abdollahi1, Sajad Ashouri2, Mohsen Abedi3
1Department of Industrial and Systems Engineering, Rochester Institute of Technology, Rochester, NY 14623, USA.
Sensors (Basel, Switzerland)
|July 2, 2020
Summary
This study developed a machine learning model using wearable sensors to classify patients with nonspecific low back pain (NSLBP) into risk groups. Kinematic data achieved ~75% accuracy, offering a quantitative approach for personalized rehabilitation.
Area of Science:
- Biomedical Engineering
- Rehabilitation Science
- Data Science
Background:
- Nonspecific low back pain (NSLBP) is a global health issue with significant socioeconomic impact.
- Current NSLBP patient categorization relies on subjective methods like the STarT Back Screening Tool (SBST).
- Objective, quantitative measures are needed for improved NSLBP assessment and personalized treatment.
Purpose of the Study:
- To develop a sensor-based machine learning model for classifying NSLBP patients.
- To utilize quantitative kinematic data (trunk motion, balance) alongside SBST output.
- To create a more objective and potentially cost-effective NSLBP assessment tool.
Main Methods:
- Ninety-four NSLBP patients were enrolled.
- Inertial Measurement Units (IMUs) recorded trunk kinematic data during repetitive movements on a balance board.
- Machine learning algorithms (SVM, MLP) were trained using kinematic data and SBST results as ground truth.
Main Results:
- The model successfully categorized patients into high vs. low-medium risk groups.
- Support Vector Machine (SVM) achieved approximately 75% accuracy.
- Multi-layer Perceptron (MLP) achieved approximately 60% accuracy.
- Time-scaled IMU signals demonstrated the highest predictive accuracy (~75%).
Conclusions:
- Quantitative kinematic data from wearable sensors can effectively classify NSLBP patients.
- This approach offers a promising avenue for developing objective diagnostic and prognostic tools.
- Wearable systems can facilitate personalized rehabilitation strategies for NSLBP in clinical and home settings.
