基于接尘工人职业健康监测的肺纤维化判别模型研究

The Discriminant Model for Pulmonary Fibrosis in Dust Exposed Workers Based on Monitoring Data of Occupational Health

目的通过分析接尘工人肺纤维化影响因素,建立接尘工人肺纤维化的Bayes判别模型,为动态监测接尘工人肺纤维化提供参考。方法采用2015-10/2017-05四川省重点职业病职业健康检查监测数据,提取其中接触煤尘有害因素的职业健康体检信息进行分析,采用Logistic回归分析筛选接尘工人肺纤维化的影响因素,构建Bayes判别模型,对接尘工人肺纤维化进行预测。结果根据监测数据,Logistic回归筛选出影响接尘工人肺纤维化有统计学意义的因素有:接触所监测危害因素工龄、血压舒张压、肺功能FEV1/FVC、心电图、企业类型、行业、年龄,综合专业知识、专家意见及判别分析特点,将总工龄、肺功能FVC、肺功能FEV1也纳入Bayes判别模型。模型经回代检验,构建的Bayes判别模型预测准确率为71.1%。ROC曲线下面积为0.801(P<0.05)。结论 Bayes判别模型可以作为动态监测接尘工人肺纤维化的辅助方法,提供划定重点监护人群的参数信息,为早期预测接尘工人肺纤维化提供参考。

Objective To analyze the influencing factors and establish Bayes discriminant model so as to provide reference for dynamic monitoring of lung fibrosis in workers exposed to dust. Methods The monitoring data of occupational health examination for key occupational diseases in Sichuan Province from October 2015 to May 2017 were used to analyze the harmful factors by Logistic regression and construct a Bayesian discriminant model to predict lung fibrosis in workers exposed to dust. Results According to the monitoring data,Logistic regression analysis showed that statistically significant factors affecting lung fibrosis in dust exposed workers included length of service exposed to the hazards（ years）,diastolic blood pressure,lung function FEV1/FVC,ECG,the enterprise type,profession and age.Considering expertise, expert opinions and the characteristics of discriminative analysis, the total length of service,pulmonary function FVC and FEV1 were also included in the Bayes discriminant model.According to return test,the prediction accuracy was 73. 8 %. The area under the ROC curve was 0. 801（ P〈0. 05） and statistically significant. Conclusion The Bayes discriminant model can be used as an auxiliary method for dynamic monitoring of pulmonary fibrosis in dust exposed workers,and provide information about the parameters of the key care population and reference for early prediction of lung fibrosis in dust exposed workers.