摘要:目的 结合大量临床诊断病例的验证分析，寻找术前甲状腺结节相关指标，建立甲状腺结节良恶性鉴别早期辅助诊断模型。方法 研究解放军总医院收治甲状腺结节住院患者并及时收集其术前血液样本用于生化、甲功、血常规等相关指标测定。回顾性验证分析术前病人一般资料、临床血液学数据、超声结果与术后病理诊断的相关性。结果 入选Logistic回归方程指标包括年龄(Age)、血清促甲状腺激素(TSH)、血清高密度脂蛋白(HDL)、中性粒细胞计数(N)、超声结果(Ult)5项，建模方程为Logit(P)= 1.673-0.069X1+0.301X3-1.499X8+ 5.335X10+21.182X12，模型组联合ROC曲线下面积为0.882；验证组为0.812。结论 通过模型分析初步建立甲状腺结节术前良恶性鉴别的血液学指标应用基础，联合超声为手术选择提供客观依据。

Abstract: Objective To establish an auxiliary diagnosis model for differentiate diagnosis of benign and malignant thyroid nodules with a great quantity of clinical case validation. Methods The serum samples of inpatients with thyroid nodules before surgery were collected for biochemistry, thyroid hormones and blood cells testing and case history were used for retrospective analysis to determine the correlation between clinical blood, ultrasound and clinicopathologic analysis results. Results Index of Age, TSH, HDL, N and Ult were selected in Logistic regression equation establishment. The ROC area under curve of model team was 0.882, and the ROC area under curve of validation team was 0.812. Conclusion The model established in this study has been preliminary applied the blood tests for differentiate diagnosis of benign and malignant thyroid nodules before surgery. It could provide objective guidance for surgery combined with ultrasonic examination.