应用判别分析模型预测女性盆腔肿物性质的研究

目的探讨依据患者肿瘤标志物及一般资料建立的判别预测模型对于盆腔肿物性质诊断的价值。方法回顾性分析124例因盆腔肿物住院接受手术的患者，依据术后病理分为卵巢良性肿物组85例，子宫平滑肌瘤组25例，卵巢恶性肿瘤组14例，将所有的数据随机分成两个数据集，分别为训练集104例和验证集20例。收集上述所有患者的一般资料并检测其肿瘤标志物水平，建立判别预测模型并对其进行验证。结果本研究将训练集通过以下变量建立了判别预测模型-1: 年龄、孕次、产次、身高、体重、BMI、初潮年龄、绝经与否、CA153、CA125、AFP、CEA、CYFRA21-1、SCCAg、CA199、抑制素B，通过交叉验证得到其准确率为91.3%。同时，将上述数据采用Logistic回归分析，寻找有统计学意义的变量，发现年龄、孕次、抑制素B、CA125等统计资料差异有统计学意义(P<0.05)。再将上述变量建立判别预测模型-2，通过交叉验证后得到其准确率为93.3%。最后应用验证集将两个模型进行验证。结论判别预测模型-1和判别预测模型-2均可用于盆腔肿物性质的预测，但后者对盆腔肿物性质的预测具有数据简化、灵敏度高等优势，对于术前临床辅助诊断具有重要的临床意义。

Construction of a prediction model for differentiating female pelvic masses

ObjectiveTo construct and verify a prediction model for differentiating female pelvic masses. MethodsClinical data of 124 patients with pelvic masses who undergoing surgery in Tsinghua Beijing Changgun Hospital from March to December 2020 were retrospectively analyzed. According to postoperative pathology, there were 85 cases of ovarian benign neoplasms, 25 cases of uterine fibroids, 14 cases of ovarian malignant tumor. Patients were randomly assigned in training set(n=104) and validation set(n=20). The factors correlated with three groups of disease were analyzed with Logistic regression, based on which a prediction model was constructed and verified. ResultsIn the training set, the prediction model 1 was established based on age, pregnancy, parity, body mass index, menarche age, menopause, CA153, CA125, AFP, CEA, CYFRA21-1, SCCAg, CA199 and inhibin B. The cross validation showed that the prediction accuracy of the model 1 was 91.3%. Logistic regression analysis showed that the age, pregnancy times, inhibin B and CA125 were significantly different among three groups(P<0.05), based on which the prediction model 2 was established. The cross validation the accuracy of model 2 was 93.3%. In validation set the prediction accuracy for model 1 and model 2 was 70%(14/20) and 80%(16/20), respectively. ConclusionThe constructed prediction model 1 and 2 in this study may be used for preoperative auxiliary diagnosis in female patients with pelvic masses, and model 2 has advantages of using less parameters and better sensitivity.