【摘要】 目的：探讨决策树模型对甲状腺乳头状癌（Papillary Thyroid Carcinoma,PTC）合并桥本氏甲状腺炎（Hashimoto’s Thyroiditis,HT）的诊断价值。方法：回顾性分析446例经手术病理证实为PTC的患者资料，其中合并HT的PTC结节127例，单纯PTC结节319例。收集结节的临床病理特征和超声特征，单因素分析筛选出有统计学意义的指标构建决策树模型，并使用受试者工作特征（Receiver Operating Characteristic,ROC）曲线验证其诊断效能。结果：决策树模型结果显示，女性，无包膜外侵犯，纵横比>1，边缘不光整，多发病灶，最大径≤1cm与PTC合并HT显著相关；决策树模型训练组准确度为84.10%，验证组准确度为80.90%,ROC曲线下面积（Area Under Curve, AUC）为0.92 (95%CI:0.89～0.94)。结论：本研究基于临床及超声特征成功构建了决策树模型，该模型对鉴别诊断HT背景下的PTC结节与单纯PTC结节具有一定的应用价值，并为临床提供一种新的诊断思路。更多还原

【Abstract】 Objective:To explore the value of decision tree model in the diagnosis of Hashimoto’s thyroiditis(HT) complicated with papillary thyroid carcinoma(PTC). Methods:The data of 446 patients with PTC confirmed by operation and pathology were analyzed retrospectively,including 127 cases with HT nodules and 319 cases with simple PTC nodules.The clinicopathological features and ultrasonic features of nodules were collected,and the statistically significant indicators were screened out by univariate analysis to construct a decision tree model.The diagnostic efficiency was verified by using the ROC curve.Results:The results of the decision tree model showed that female,had no extracapsular invasion,tall to wide ratio>1,uneven edge,multiple lesions,and the largest diameter ≤1cm was significantly related to PTC with HT. The accuracy of the decision tree model training group was 84.10%, and the accuracy of the verification group was 80.90%.The area under the curve(AUC) was 0.92(95%CI: 0.89 to 0.94).Conclusion:Based on the clinical and ultrasonic features,the decision tree model has been successfully constructed in the study.The model has certain application value in differentiating PTC nodules from simple PTC nodules in HT background, and provides a new diagnostic idea for clinic.更多还原