温石棉矿工人消化道肿瘤的贝叶斯暴露-反应模型  

Bayesian exposure-response model of digestive tract tumors in chrysotile workers

作　　者：陈耿 黄倩 张杰 兰亚佳[2] CHEN Geng;HUANG Qian;ZHANG Jie;LAN Ya-jia(School of Computer Science,Chengdu Normal University,Chengdu,Sichuan 611130,China;不详)

机构地区：[1]成都师范学院计算机科学学院,四川成都611130 [2]四川大学华西公共卫生学院/华西第四医院,四川成都610041 [3]成都师范学院智能软件现代产业学院,四川成都611130

出　　处：《现代预防医学》2023年第19期3500-3505,共6页Modern Preventive Medicine

基　　金：职业肿瘤与职业性听力损失的风险评估技术和生物标志物研究(2022YFC2503203);成都师范学院课程思政项目(XJKCSZKC2026数据结构与算法设计)。

摘　　要：目的 采用迭代重加权最小二乘法(IRLS)和贝叶斯方法求解温石棉矿工人消化道肿瘤暴露-反应模型,为温石棉从业人员健康防护提供依据。方法 使用1981—2014年对中国西部某温石棉矿1 929工人队列数据为训练集,分别以IRLS和贝叶斯方法求解基于泊松广义线性模型的温石棉矿工人消化道肿瘤的暴露-反应模型。结果 把观测数据随机划分为训练集和验证集,分别用IRLS和贝叶斯方法求解模型,贝叶斯方法求解的模型在验证集上的平均绝对误差值更小。模型预测显示:温石棉暴露浓度与消化道肿瘤人数呈显著正相关性(β的均值:1.500,95%HDI:0.927~2.122),且随着暴露浓度加大消化道肿瘤人数的增长速率也在上升。结论 在数据量少的情况下,贝叶斯方法比IRLS方法更适合于求解此温石棉矿工人消化道肿瘤暴露-反应模型。本研究支持温石棉暴露可增加消化道肿瘤风险,此模型可为温石棉从业人员的消化道肿瘤风险监测和防控提供数值参考。Objective To solve the exposure-response model of digestive tract tumor of chrysotile workers by iterative reweighted least square method(IRLS)and Bayesian method,to provide reference for health protection of chrysotile workers.Methods Using the cohort data of 1929 workers in a chrysotile asbestos mine in western China from 1981 to 2014 as the training set,IRLS and Bayesian methods were used to solve the exposure-response model of gastrointestinal tumors of chrysotile workers based on Poisson generalized linear model.Results The observed data were randomly divided into training set and validation set,and the model was solved by IRLS and Bayesian method,respectively.The average absolute error of the model solved by Bayesian method was smaller.The model prediction showed that the exposure concentration of chrysotile asbestos had a significant positive correlation with the number of digestive tract tumors(the average value of chrysotile asbestos exposure was 1.500,95%HDI:0.927-2.122),and the growth rate of the number of digestive tract tumors increased with the increase of exposure concentration.Conclusion In the case of small amount of data,Bayesian method is more suitable than IRLS method to solve the digestive tract tumor exposure-response model of chrysotile workers.This study supports that chrysotile asbestos exposure can increase the risk of digestive tract tumors,and this model can provide numerical reference for the risk monitoring and prevention of digestive tract tumors in chrysotile workers.

关 键 词：贝叶斯分析 温石棉 职业队列 消化道肿瘤 暴露反应 

分 类 号：R136.3[医药卫生—劳动卫生] R735[医药卫生—公共卫生与预防医学]