# Determinants of work-related musculoskeletal disorders among coal miners in Jining, China: development of a predictive risk model

## TL;DR

This study identifies risk factors for musculoskeletal disorders in Chinese coal miners and builds a machine learning model to predict these risks.

## Contribution

The novel contribution is the development of an ensemble machine learning model for predicting WMSD in coal miners using personal and work-related factors.

## Key findings

- The 12-month prevalence of WMSD among coal miners was 82%, with neck, shoulders, and lower back most affected.
- Smoking, health perception, and uncomfortable postures were significant risk factors (p < 0.05).
- A neural network model achieved an AUC of 0.886 on training data, and a fused model outperformed individual models.

## Abstract

Work-related musculoskeletal disorders (WMSD) are highly prevalent among coal miners and pose a significant threat to occupational health. Understanding the underlying risk factors and developing a predictive model for WMSD risk can help to mitigate WMSD.

To identify key determinants of WMSD among coal miners in Jinang, China, and construct a predictive model to assess risk.

One thousand four hundred nine coal miners from two coal mining companies were surveyed using the modified Chinese Muscle Questionnaire (CMQ). Prevalence rates and risk factors were assessed using logistic regression. Machine learning algorithms were applied to construct the predictive model.

The 12-month overall prevalence of WMSD was 82%, with the neck (59.5%), shoulders (53.4%), and lower back (46.5%) being the most affected. Eight variables, including smoking behaviors, perceived health status, and uncomfortable working posture, were significantly associated with WMSD (p < 0.05). The neural network model achieved the highest performance (area under the curve: 0.886 on training and 0.704 on test). The fused model outperformed individual models in the final stacking integration learning.

Work-related musculoskeletal disorders are highly prevalent among Chinese coal miners and are influenced by personal and work-related factors. Machine learning models, particularly ensemble approaches, offer promise for risk prediction and targeted prevention.