张雨晴, 许宁, 武云云, 王强. 应用随机森林分析非吸烟女性肺癌风险因素[J]. 环境卫生学杂志, 2022, 12(2): 80-86. DOI: 10.13421/j.cnki.hjwsxzz.2022.02.002
    引用本文: 张雨晴, 许宁, 武云云, 王强. 应用随机森林分析非吸烟女性肺癌风险因素[J]. 环境卫生学杂志, 2022, 12(2): 80-86. DOI: 10.13421/j.cnki.hjwsxzz.2022.02.002
    ZHANG Yu-qing, XU Ning, WU Yun-yun, WANG Qiang. Using Random Forest to Analyze Risk Factors of Female Nonsmoking Lung CancerUsing random forest method to analyze risk factors for lung cancer among non-smoking women[J]. Journal of Environmental Hygiene, 2022, 12(2): 80-86. DOI: 10.13421/j.cnki.hjwsxzz.2022.02.002
    Citation: ZHANG Yu-qing, XU Ning, WU Yun-yun, WANG Qiang. Using Random Forest to Analyze Risk Factors of Female Nonsmoking Lung CancerUsing random forest method to analyze risk factors for lung cancer among non-smoking women[J]. Journal of Environmental Hygiene, 2022, 12(2): 80-86. DOI: 10.13421/j.cnki.hjwsxzz.2022.02.002

    应用随机森林分析非吸烟女性肺癌风险因素

    Using Random Forest to Analyze Risk Factors of Female Nonsmoking Lung CancerUsing random forest method to analyze risk factors for lung cancer among non-smoking women

    • 摘要:
      目的 应用随机森林探索非吸烟女性肺癌风险因素,为非吸烟女性肺癌的预防提供更多参考。
      方法 2017年8月-2021年8月,在辽宁、江苏、安徽及青海省,进行以医院为基础的多中心成组病例对照研究,共纳入1 315名研究对象,以三级及三级以上医院确诊的非吸烟女性肺癌新发病例为病例组(679人),以同家医院就诊或体检的女性非癌症患者或健康女性为对照组(636人),采用以亚洲女性肺癌联盟(FLCCA)和宣威女性肺癌问卷为基础的调查问卷,收集一般人口统计学特征、暴露因素、行为因素、既往史、家族史等数据,利用随机森林模型和多因素Logistic回归分析非吸烟女性肺癌风险因素。
      结果 最终纳入1 294名研究对象,均为女性,经χ2检验和U检验,病例组(n=668)和对照组(n=626)年龄、民族、文化程度差异无统计学意义(P>0.05)。经5次重复10折交叉验证,经随机森林分析,被动吸烟、住在地下室或一楼满二十年、住宅附近存在冶炼厂满二十年、慢性支气管炎史、母亲孕期吸烟、一级亲属肺癌家族史、肺炎史、职业辐射暴露满十年、职业农药暴露和其他化学物职业暴露满十年10个因素依次为非吸烟女性肺癌主要危险因素。经Logistic回归分析,被动吸烟、住在地下室或者一楼满二十年及住宅5 km范围内存在冶炼厂满二十年是非吸烟女性肺癌的危险因素,OR(95%CI)值分别为1.42(1.07~1.89)、1.56(1.17~2.10)和2.27(1.19~4.56)。一级亲属家族史、既往肺炎和化学物质职业暴露满十年与非吸烟肺癌风险关联边缘显著(P < 0.10)。
      结论 除室内外空气污染外,肺部炎症和化学物质的职业暴露可能也是非吸烟女性肺癌需要关注的风险因子。

       

      Abstract:
      Objective To explore risk factors for lung cancer among non-smoking women by using the random forest method, and provide a reference for preventing lung cancer among non-smoking women.
      Methods From August 2017 to August 2021, a hospital-based multicenter case-control study was conducted, involving 1 315 participants from Liaoning, Jiangsu, Anhui, and Qinghai province, with 679 non-smoking women with newly diagnosed lung cancer at tertiary or higher-level hospitals in the case group and 636 non-cancer female patients or healthy women who sought care or underwent physical examination at the same hospitals in the control group. A revised questionnaire based on Female Lung Cancer Consortium in Asia and Xuanwei Female Lung Cancer Questionnaire was adopted to collect information on demography, exposure factors, behavioral factors, past history and family history. The random forest model and multivariate logistic regression analysis were used to analyze risk factors for lung cancer among non-smoking women.
      Results A total of 1 294 female participants were finally included in analysis. There was no significant difference in age, nation, and education level (P>0.05) between the case group (n=668) and the control group (n=626). After running 10-fold cross-validation 5 times, the random forest analysis showed that the following ten factors were the main risk factors for lung cancer in non-smoking women: passive smoking, living on the ground floor or in the basement for at least 20 years, the presence of a smelter plant near residence for at least 20 years, a history of chronic bronchitis, maternal smoking during pregnancy, a family history of lung cancer of first-degree relatives, a history of pneumonia, and occupational exposure to radiation, pesticides, or other chemicals for at least 10 years. The logistic regression analysis showed that passive smoking (OR: 1.42, 95%CI: 1.07-1.89), living on the ground floor or in the basement for at least 20 years (OR: 1.56, 95%CI: 1.17-2.10), and the presence of a smelter plant within 5 km around residence for at least 20 years (OR: 2.27, 95%CI: 1.19-4.56) were the risk factors for lung cancer among non-smoking women; a family history of lung cancer of first-degree relatives, a history of pneumonia and occupational exposure to chemicals for at least 10 years were marginally associated with the risk of lung cancer among non-smoking women (P < 0.10).
      Conclusion Besides exposure to indoor and outdoor air pollution, pneumonia and occupational exposure to chemicals may also be risk factors for lung cancer in non-smoking women.

       

    /

    返回文章
    返回