A Predictive Model for Abnormal Bone Density in Male Underground Coal Mine Workers.

Area of Science:

• Occupational Health
• Data Science in Medicine
• Skeletal Health Research

Background:

• Underground coal mining environments pose significant risks to skeletal health.
• Early identification of abnormal bone density is crucial for miner protection.
• Developing accurate risk prediction models is essential for occupational health interventions.

Purpose of the Study:

• To develop and compare the predictive performance of Logistic Regression, Convolutional Neural Network (CNN), and XG Boost models for abnormal bone density in coal miners.
• To identify the optimal model for early risk assessment of skeletal health issues in this population.

Main Methods:

• A cohort of 3695 male underground coal miners was analyzed.
• Predictor variables were identified via single-factor analysis and literature review.
• Logistic Regression, CNN, and XG Boost models were developed and evaluated using sensitivity, specificity, F1 score, Brier score, and AUC.

Main Results:

• The XG Boost model demonstrated superior performance across all evaluated metrics (sensitivity, specificity, F1 score, Brier score, AUC) compared to Logistic Regression and CNN.
• XG Boost achieved the highest F1 score (0.919) and the lowest Brier score (0.040) and Calibration-in-the-large (0.020).
• The superior predictive accuracy of the XG Boost model was consistent across training, testing, and validation datasets.

Conclusions:

• The XG Boost model exhibits high practical value for predicting abnormal bone density in underground coal miners.
• This model significantly outperforms traditional Logistic Regression and CNN in identifying miners at risk.
• Implementing the XG Boost model can enhance skeletal health surveillance and preventative strategies in the coal mining industry.