Classification of musculoskeletal pain using machine learning

Area of Science:

• Biomedical Engineering
• Computational Health Science
• Data Science in Healthcare

Background:

• Musculoskeletal pain is a widespread health issue impacting productivity and quality of life.
• Identifying risk factors for musculoskeletal pain is crucial for developing effective interventions.
• Existing assessment methods may benefit from advanced computational approaches.

Purpose of the Study:

• To develop and evaluate a framework using Particle Swarm Optimization (PSO) integrated with neural networks for musculoskeletal pain risk assessment.
• To identify key demographic, professional, physical, and lifestyle factors contributing to musculoskeletal pain.
• To enhance the accuracy and efficiency of musculoskeletal pain risk detection.

Main Methods:

• A comprehensive dataset of 350 individuals including pain experiences, demographics, profession, physical, and lifestyle data was utilized.
• Data preprocessing involved handling missing values, class balancing with SMOTE, and feature normalization.
• A feedforward neural network optimized by PSO was employed to predict pain risk.
• Performance was evaluated using accuracy, precision, recall, F1-score, and AUC-ROC.

Main Results:

• The PSO-optimized neural network achieved high performance in identifying musculoskeletal pain risk (95.8-100% accuracy).
• Significant risk determinants identified include age, BMI, exercise frequency, and occupational factors.
• The framework demonstrated superior performance compared to conventional assessment methods.

Conclusions:

• The proposed PSO-integrated neural network framework is effective for musculoskeletal pain risk assessment.
• The identified risk factors provide valuable insights for developing targeted preventive strategies.
• Optimization techniques show significant potential for advancing musculoskeletal pain assessment and prevention.