基于健康体检的队列研究构建肺癌风险预测模型  

Construction of lung cancer risk prediction model based on a cohort study of physical examination

作　　者：迪丽热巴·艾斯卡尔 高艳 张莉[2] Dilireba Aisikaer;GAO Yan;ZHANG Li(College of Health Management,Xinjiang Medical University,Urumqi 830017,China;Cadre Health Care Center,the First Affiliated Hospital of Xinjiang Medical University,Urumqi 830054,China)

机构地区：[1]新疆医科大学健康管理学院,乌鲁木齐830017 [2]新疆医科大学第一附属医院干部保健中心,乌鲁木齐830054

出　　处：《新疆医科大学学报》2025年第3期368-376,共9页Journal of Xinjiang Medical University

基　　金：新疆维吾尔自治区卫生健康保健科研专项项目(BL202419)。

摘　　要：目的基于一项前瞻性健康体检筛查队列,评估肺癌发生的风险因素,以此构建列线图风险预测模型。方法选择新疆某三甲医院2019年1月1日至12月31日进行低剂量螺旋CT检查并查出肺结节的2607人,最终纳入研究的有2504人。收集临床特征及实验室检查结果,并开展问卷调查,随访至2023年12月31日,观察肺癌的发病率。利用单因素比较和多因素Cox回归分析风险因素,并构建列线图建模。使用受试者工作特征(ROC)曲线和曲线下面积(AUC)评价模型。通过校准曲线评估模型预测的风险与实际观察到的风险之间的一致性,决策曲线分析(DCA)检验预测模型的临床效益。结果性别、在职状态、吸烟史、癌症既往病史、慢性呼吸道疾病史、精神压力、糖类抗原CA72-4、鳞状细胞癌相关抗原为肺癌的风险因素,将此8个指标纳入列线图预测模型,在建模组和验证组中均表现出良好的效能,曲线下面积分别为0.954、0.942、0.915、0.872、0.914和0.900。校准曲线提示,预测的肺癌发生概率和实际观测概率一致性较高。决策曲线(DCA)显示模型有良好的临床适用性。结论模型在建模组和验证组中均展现出良好效能,其曲线下面积可观,本研究所建立的模型对新疆地区健康人群发生肺癌的风险有预测价值。Objective Based on a prospective physical examination screening cohort,the risk factors of lung cancer were evaluated,and the nomogram wind prediction model was constructed.Methods 2607 people who underwent low-dose spiral CT examinations and were detected with pulmonary nodules in the hospital from January 1,2019 to December 31,2019 were screened.A total of 2504 individuals were ultimately included in the model.The clinical characteristics and laboratory test results of the research subjects were collected and questionnaire survey was conducted.Followup was conducted until December 31,2023 to observe the incidence of lung cancer.Univariate comparisons and multivariate Cox regression analyses were used to analyze the risk factors,and nomogram modeling was conducted.The receiver operating characteristic(ROC)curve and the area under the curve(AUC)were used to evaluate the model.The consistency between the predicted risk by the model and the actually observed risk was evaluated through calibration curves,and the clinical benefit of the prediction model was examined by decision curve analysis(DCA).Results Gender,employment status,smoking history,previous history of cancer,history of chronic respiratory diseases,mental stress,carbohydrate antigen CA72-4 and squamous cell carcinoma-related antigen were the risk factors for lung cancer.These 8 indicators were incorporated into the nomogram prediction model,which demonstrated good efficacy in both the training and validation sets.The areas under the curves were 0.954,0.942,0.915,0.872,0.914 and 0.900,respectively.The calibration curves indicated a high consistency between the predicted probability of lung cancer occurrence and the actually observed probability.The decision curve(DCA)demonstrated that the model had good clinical applicability.Conclusion The model showed good performance in both the modeling and validation groups,with a considerable area under the curve.The model established in this study has predictive value for the risk of lung cancer in healthy populati

关 键 词：肺恶性肿瘤 早期筛查 风险因素 预测模型 健康体检 

分 类 号：R184[医药卫生—流行病学]