【摘要】 目的 挖掘肺癌远处转移患者人群预后因素并建模，帮助该患病人群提升预后质量，辅助临床诊疗。方法 下载2010-2015年美国国立癌症研究所的监测、流行病学和结果（SEER）数据库中肺癌患者临床病理学、人口统计学、治疗及辅助检查与评价信息，应用PSM分组技术，筛选出目标研究对象，应用Cox单因素和多因素分析，筛选肺癌远处转移患者人群预后因素。按7∶3分为训练组与验证组，据此构建Nomogram预测模型实现因子应用价值，应用受试者操作特征（ROC）曲线、校准曲线评估模型可靠性，应用决策曲线分析（DCA）评估模型的临床应用价值，风险分层研究评价模型应用意义，KM评价风险分层结果。结果 PSM分组技术使得各项基线特征SMD中，有87.5%小于或在0.1附近（0.015以内），数据偏倚性得到了有效消除；24个预测因子筛选结果覆盖全面，其中淋巴领域预测因子表现突出，首次加入评价体系字段进入肺癌远处转移患者预后研究；ROC曲线提示模型预测结果为0.781～0.815，具有较好区分度，且校准曲线显示预测模型与理想模型拟合良好，DCA显示出临床净获益良好，1～5年生存率区间可覆盖（0.24～0.84）；风险分层效果显著，P<0.0001,KM结果具有统计学意义。结论 基于临床病理学、人口统计学、治疗及辅助检查与评价信息因素构建的肺癌远处转移多维度综合性Nomogram模型及风险分层具有可靠的预测性能及临床实用价值，对临床上实施个体化预测生存时间及制定诊疗计划、护理方案具有重要参考意义。更多还原

【Abstract】 Objective To explore and model the prognostic factors of patients with distant metastasis of lung cancer, so as to help the patients improve the quality of prognosis and assist clinical diagnosis and treatment. Methods The clinicopathological, demographic, treatment and auxiliary examination and evaluation information of lung cancer patients in the Surveillance, Epidemiology, and End Results(SEER) database of the National Cancer Institute from 2010 to 2015 were downloaded. PSM grouping technique was used to screen out the target subjects. Cox univariate and multivariate analysis were used to screen the prognostic factors of lung cancer patients with distant metastasis. According to 7 ∶3, they were divided into the training group and the verification group. Based on this, the Nomogram prediction model was constructed to realize the application value of the factors. The receiver operator characteristic(ROC) curve and calibration curve were used to evaluate the reliability of the model. The decision curve analysis(DCA) was used to evaluate the clinical application value of the model, the risk stratification study was used to evaluate the application significance of the model, and KM was used to evaluate the risk stratification results. Results PSM grouping technology made 87.5% of the SMD of each baseline feature less than or within 0.1(within 0.015), and the data bias was effectively eliminated. The screening results of 24 predictors were comprehensive, among which the predictive factors in the lymphatic field were outstanding. For the first time, the evaluation system field was added into the prognosis study of patients with distant metastasis of lung cancer. The ROC curve suggested that the prediction results of the model could were 0.781-0.815, with good discrimination, and the calibration curve showed that the prediction model fitted well with the ideal model. DCA showed good clinical net benefit, and the 1-5-year survival rate interval could be covered(0.24-0.84). The effect of risk stratification was significant, P<0.0001, and the KM results were statistically significant. Conclusion The multi-dimensional comprehensive Nomogram model and risk stratification of distant metastasis of lung cancer based on clinicopathological, demographic, treatment and auxiliary examination and evaluation information factors have reliable predictive performance and clinical practical value. It has important reference significance for clinical implementation of individualized prediction of survival time and formulation of diagnosis and treatment plan and nursing plan.更多还原