构建胰腺癌术后复发预测模型：基于SEER数据库
Constructing a Predictive Model for Pancreatic Cancer Postoperative Recurrence: Based on the SEER Database

目的：发现关于胰腺癌术后复发风险要素，构建胰腺癌患者术后复发的风险预测模型。方法：从SEER数据库中提取2000年至2018年期间诊断为胰腺癌的3960例患者的临床资料，随机分为建模组和验证组，使用单因素和多因素Logistic回归模型分析胰腺癌患者术后复发的风险因素并构建列线图模型；使用ROC曲线及的曲线下面积(AUC)和校正曲线评估模型的准确性，使用DAC决策曲线评估患者的获益程度。结果：多因素Logistic回归分析发现，年龄、原发灶部位、病理类型、组织学分级、T分级、淋巴结阳性个数、放化疗情况是患者预后的独立影响因素(P < 0.05)。训练组ROC曲线下面积(AUC值)为0.765 (95%CI: 0.746~0.783)，验证组ROC曲线下面积(AUC值)为0.770 (95%CI: 0.742~0.798)，校准曲线表明预测值与实际值有良好的一致性，决策曲线(DCA)显示模型有良好的临床效用。结论：本研究成功构建了胰腺癌患者术后复发的风险预测模型。
Objective: To find out the risk factors of postoperative recurrence of pancreatic cancer and construct a risk prediction model for postoperative recurrence of pancreatic cancer. Methods: The clinical data of 3960 patients diagnosed with pancreatic cancer from 2000 to 2018 were extracted from SEER database and randomly divided into a modeling group and a validation group. Single factor and multi factor logistic regression models were used to analyze the risk factors of postoperative recurrence of pancreatic cancer patients and construct a nomogram model; ROC curve and area under curve (AUC) and calibration curve were used to evaluate the accuracy of the model, and DAC decision curve was used to evaluate the degree of benefit for patients. Results: Multivariate logistic regression analysis revealed that age, primary lesion location, pathological type, histological grade, T-grade, number of positive lymph nodes, and radiation and chemotherapy status were independent prognostic factors for patients (P < 0.05). The area under the ROC curve (AUC value) of the training group was 0.765 (95%CI: 0.746~0.783), while the area under the ROC curve (AUC value) of the validation group was 0.770 (95%CI: 0.742~0.798). The calibration curve showed good consistency between the predicted and actual values, and the decision curve (DCA) showed that the model had good clinical utility. Conclusion: This study successfully constructed a risk prediction model for postoperative recurrence of pancreatic cancer.