基于LightGBM的煤矿从业人员肺通气功能障碍风险预测OA

To prevent and control pulmonary ventilation dysfunction in coal mine workers and reduce losses caused by coal workers'pneumoconiosis,this study used occupational health examination data of coal mine workers to screen eight important indicators affecting pulmonary ventilation dysfunction.A prediction model for pulmonary ventilation dysfunction in coal mine workers was constructed based on the LightGBM algorithm,and its hyperparameters were optimized by grid search.Prediction accuracy,precision,recall,F1 score,and AUC were used to comprehensively evaluate the model,and the results were compared with those of LR and XGBoost,two commonly used machine learning algorithms.The results show that,among all influencing factors,age has the greatest effect on pulmonary ventilation dys-function,followed by smoking and BMI.The LightGBM-based model achieves the best predictive per-formance,with an AUC of 0.85,prediction accuracy of 81%,precision of 79%,recall of 81%,and an F1 score of 0.77,all higher than those of the XGBoost and LR models.Among the 79 coal mine work-ers included in the risk prediction of pulmonary ventilation dysfunction,8 were identified as possibly having the disorder.Coal mine workers aged 30～39 years,with 10～15 years of service,exposed to dust in their work,with overweight or obese BMI,and who regularly smoke and drink have a higher risk of pulmonary ventilation dysfunction.The study provides a reference for workers'health protection and the stable development of enterprises.