目的：基于监测、流行病学和最终结果（surveillance, epidemiology, and end results, SEER）数据库，分析影响胰腺癌预后的独立因素并构建预测模型。方法：本研究从SEER数据库获取2010—2015年美国7 801例胰腺癌病人的临床资料，以7∶3的比例随机分为建模组、验证组。对建模组临床变量进行多因素COX回归分析获得影响生存的独立因素，构建列线图。通过受试者操作特性（receiver operating characteristic, ROC）曲线和校准曲线验证模型的准确性。结果：年龄、原发部位、病理分级、T分期、N分期、M分期、手术方式、放疗、化疗与胰腺癌的预后相关，总生存的3年、5年ROC曲线下面积（area under cure, AUC）分别为0.90、0.91，癌症特异性生存分别为0.91、0.91。校准曲线显示观察值与预测值之间具有良好的一致性。经筛选得到的临床变量确实对胰腺癌预后有影响。结论：所构建的模型具有较好的预测准确性，有助于胰腺癌病人的临床决策和个性化治疗。

Objective To analyze the independent factors affecting the prognosis of pancreatic cancer and construct a prediction model based on surveillance, epidemiology, and end results (SEER) database. Methods The clinical data of 7 801 American pancreatic cancer patients from 2010 to 2015 were obtained from SEER database. They were randomly divided into training group and validation group in a ratio of 7:3. The nomogram was constructed after multivariate COX regression analysis of clinical variables in the training group. The accuracy of the model was verified by receiver operating characteristic(ROC) curve and calibration curve. Results Age, primary-site, grade, T-stage, N-stage, M-stage，surgery, radiotherapy and chemotherapy were related to the prognosis of the patients with pancreatic cancer. The area under curve(AUC) of overall survival(OS) ROC curve of 3- and 5-year were 0.90 and 0.91 respectively. The AUC of cancer specific survival(CSS) ROC curve were 0.91 and 0.91 respectively. The calibration curve showed a good consistency between the observed and predicted values. The selected clinical variables did have an impact on the prognosis of the patients with pancreatic cancer. Conclusions The model had good prediction accuracy and was helpful for clinical decision-making and personalized treatment of the patients with pancreatic cancer.