Related Experiment Video
Updated: May 26, 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
404
Biomechanical Risk Classification in Repetitive Lifting Using Multi-Sensor Electromyography Data, Revised National
Fatemeh Davoudi Kakhki1,2, Hardik Vora1,3, Armin Moghadam4
1Machine Learning & Safety Analytics Lab, School of Engineering, Santa Clara University, Santa Clara, CA 95053, USA.
Biosensors
|February 25, 2025
Summary
This study uses wearable sensors and deep learning to accurately assess repetitive lifting risks in manufacturing, improving workplace safety and reducing injuries. The Convolutional Neural Networks model demonstrated high precision in identifying high-risk tasks.
Area of Science:
- Occupational Ergonomics
- Biomechanical Risk Assessment
- Machine Learning in Safety
Background:
- Repetitive lifting tasks in manufacturing commonly lead to shoulder injuries, affecting worker health and productivity.
- Current methods for assessing biomechanical risk are often subjective and lack accuracy.
- Accurate, real-time risk assessment is crucial for preventing musculoskeletal disorders.
Purpose of the Study:
- To develop and validate deep learning models for classifying occupational lifting risk using electromyography (EMG) data.
- To create a comprehensive dataset of EMG signals during repetitive lifting tasks.
- To enable precise, real-time, and dynamic risk assessments for enhanced workplace safety.
Main Methods:
- Collected time-series EMG data from 25 participants performing repetitive lifting tasks using wearable sensors.
- Calculated the lifting index using the revised National Institute for Occupational Safety and Health (NIOSH) lifting equation (RNLE).
- Developed and compared three deep learning models (CNN, MLP, LSTM) for risk classification using extracted statistical features from EMG data.
Main Results:
- The Convolutional Neural Networks (CNN) model achieved the highest performance with 98.92% precision and 98.57% recall.
- Deep learning models effectively classified lifting risk levels based on EMG data.
- Statistical features extracted from EMG data provided actionable insights for risk assessment.
Conclusions:
- Integrating wearable EMG sensors with deep learning models enables accurate, real-time occupational lifting risk assessment.
- The CNN model is highly effective for real-time risk assessment in dynamic work environments.
- This approach significantly enhances workplace safety protocols and can reduce work-related musculoskeletal disorders.
