接受体外放射治疗的非手术胆管癌患者预后因素分析及Nomogram预测模型的构建和验证.pdfVIP

目录

中文摘要············································································1

英文摘要············································································3

英文缩写············································································6

研究论文

接受体外放射治疗的非手术胆管癌患者预后因素分析及Nomogram

预测模型的构建及验证

前言············································································8

材料与方法····································································9

结果··········································································12

附图··········································································15

附表··········································································20

讨论··········································································24

结论··········································································27

个人简历···········································································51

接受体外放射治疗的非手术胆管癌患者预后因素分析及

Nomogram预测模型的构建和验证

摘要

分析接受体外放射治疗（EBRT）的非手术胆管癌（CCA）患者

预后因素，建立有效的Nomogram预测模型，对患者的预后进行个体

化预测评估。

本研究从检测、流行病和最终结果（SEER）数据库总共18个登

记中心提取合适的患者病例，按照7：3的随机分配比例，13个登记

中心的患者资料被纳入模型建立的训练队列，另外5个中心的患者登

记资料被纳入到模型的内部验证队列，对患者的信息资料进行筛选并

分析。用Kaplan-Meier方法计算中位生存时间及总生存率，log-rank

方法检验差异。为确保筛选出的独立预后变量以及预测模型更加科学

化，分别建立以下三种不同模型，在训练队列中进行预后变量筛选:

①利用单、多因素Cox回归分析后的显著意义变量（P0.05）建立变

量模型一；②逐步回归后基于最小Akaike信息准则（AIC）选择的变

量建立变量模型二；③通过10倍的交叉验证后调整参数的Lasso回归

分析，筛选出的独立预后因子建立变量模型三。最终根据筛选结果，

选择三种模型共同筛选出的变量作为最终的预后变量建立Nomogram，

预测6个月和12个月的总生存率。

同时，回顾性收集2010年1月-2020年12月空军特色医学中心

放射治疗科所收治的非手术CCA患者资料，记录年龄、性别、个人

史等患者基本信息以及患者肿瘤相关信息，包括入院时的EGCO评分、

确诊方式、肿瘤的原发部位、肿瘤分期和分化等级、有无接受化疗等。

基于本院的单中心随访数据，建立外部验证集，对建立的预测模

型进行外部验证。为了确保内部验证的有效性，使用了Bootstraps算

法，并且对内部验证队列进行了1000次的自我检查和重新抽取，利

用一致性指数（C-index）、受试者工作