LJospital医皖管理论坛MedicalManagement肺癌检测指标筛选及其在肺癌风险预测中的应用ScreeningofLungCancerDetectionIndicatorsandItsApplicationinLungCancerRiskPrediction唐添斌TANGTian-bin叶申怡YEShen-yi.②季紫薇JIZi-wei@摘要Abstract目的研究血清肺肿瘤标志物与生化、血常规指标联合检测在肺癌诊断中的价值。方法纳入肺癌患者和肺部良性病变患者各117例，收集患者治疗前的血常规、生化和血清肺肿瘤标志物的结果。将患者分为训练集和验证集，在训练集中利用Logistic回归筛选指标建立风险预测模型，绘制ROC和Calibration曲线，并在验证集中对模型进行验证。结果在肺部良性病变组中，ALT、AST、GGT、TBA、TBIL和DBIL之间呈正相关，而在肺癌组中，ALT、AST、GGT、ALP、TBIL、DBIL、LDH和CYFRA21-1之间呈正相关。通过Logistic回归筛选实验室指标并优选最佳预测模型T4-Model(CEA+CYFRA21-1+LYM%+NEU%)。其在训练集和验证集中的AUC分别为0.808和0.782。Calibration曲线显示其在训练集和验证集上具有良好的一致性。结论T4-Model对肺癌表现出较好的预测能力，可提高肺癌诊断效能。ObjectiveTostudythevalueofcombineddetectionofserumlungtumormarkers,biochemicalandbloodroutineindicatorsinthediagnosisoflungcancer.MethodsAtotalof117patientswithlungcancerand117patientswithbenignpulmonarylesionswereenrolled.Theresultsofbloodroutine,biochemicalandserumlungtumormarkerswerecollectedbeforetreatment.Thepatientsweredividedintotrainingsetandvalidationset.Logisticregressionwasusedtoestablishariskpredictionmodelinthetrainingset,ROCcurveandCalibrationcurveweredrawn,andthemodelwasverifiedinthevalidationset.ResultsPositivecorrelationswerefoundamongALT,AST,GGT,TBA,TBILandDBILinthebenignpulmonarylesiongroup,whilepositivecorrelationswerefoundamongALT,AST,GGT,ALP,TBIL,DBIL,LDHandCYFRA21-1inthelungcancergroup.ThelaboratoryindicatorswerescreenedbyLogisticregressionandthebestpredictionmodelT4-Model(CEA+CYFRA21-1+LYM%+NEU%)wasselected.ItsAUCinthetrainingsetandvalidationsetwere0.808and0.782,respectively.TheCalibrationcurveshowedgoodagreementbetweenthetrainingsetandthevalidationset.ConclusionTheT4-Modelhasagoodpredictiveabilityforlungcancerandcanimprovethediagnosticefficienc...
