探讨胃癌根治术后发生感染性并发症的危险因素，建立并验证Nomogram预测模型。

回顾性分析2010年1月至2023年4月行胃癌根治术的600例胃癌患者资料。按照7∶3的比例随机将600名患者分为训练集和内部验证集。针对训练集行LASSO回归筛选变量，通过多因素Logistic回归分析构建Nomogram模型并利用内部验证集进行内部验证，采用受试者工作特征（ROC）曲线、校准曲线、决策曲线分析（DCA）评估模型效果。所有统计分析均使用R软件(版本4.2.2)以及MSTATA软件进行。

600例行胃癌根治术的患者中，109例发生了术后感染，总感染发生例数为132例，医院感染发病率为22.0%。训练集与内部验证集患者的年龄、NRS评分、吸烟史、高血压病、糖尿病等临床特征差异无统计学意义(P>0.05)，N分期除外（P=0.18）。LASSO回归筛选出7个回归系数非零的潜在预测因子：年龄、冠心病、前白蛋白、手术时间、术中出血量、术式、T分期。多因素Logistic回归分析以上7个因素，最终得出年龄>60岁、患有冠心病、术前较低的前白蛋白、手术时间>180min、术式选择为近端胃或全胃切除术是胃癌根治术后感染的独立危险因素（P <0.05），最终基于以上因素建立Nomogram预测模型。训练集及内部验证集的ROC曲线下面积（AUC）分别为0.753、0.736，校准曲线及DCA提示模型有良好的预测能力及临床实用性，并建立了基于在线网络的交互式动态Nomogram应用程序。

本研究构建了简单且实用的胃癌根治术后感染性并发症预测工具，可以量化术后感染的个体风险，从而促进早期的预防和治疗的实施。

To investigate the risk factors of infectious complications after radical gastrectomy of gastric cancer, and establish a Nomogram prediction model.

Data of 600 patients with gastric cancer who underwent radical gastrectomy from January 2010 to April 2023 were retrospectively analyzed. The 600 patients were randomly divided into a training set and an internal validation set at a ratio of 7∶3. Based on LASSO regression screening variables of the training set, a Nomogram model was constructed by multivariate Logistic regression analysis and verified internally by internal verification set. Receiver operating characteristic (ROC) curve, calibration curve and decision curve analysis (DCA) were used to evaluate the model effect. All statistical analyses were performed using R software (version 4.2.2) and MSTATA software.

Among 600 patients undergoing radical gastrectomy, 109 cases developed postoperative infection, the total number of infections was 132 cases, and the incidence of nosocomial infection was 22.0%. There was no significant difference in clinical features such as age, NRS score, smoking history, hypertension and diabetes between the training set and the internal verification set (P >0.05), except for N stage (P =0.18). Seven potential predictors with non-zero regression coefficients were identified by LASSO regression: age, coronary heart disease, prealbumin, operative time, intraoperative blood loss, operation type, and T stage. Multivariate Logistic regression analysis of the above 7 factors showed that age >60 years old, coronary heart disease, low prealbumin before surgery, operation time >180min, proximal gastric resection or total gastrectomy were independent risk factors for postoperative infection after radical gastrectomy (P <0.05). Finally, a Nomogram prediction model was established based on the above factors. The area under ROC curve (AUC) of the training set and the internal validation set were 0.753 and 0.736, respectively. The calibration curve and DCA indicated that the model had good predictive ability and clinical practicability. An interactive Nomogram application based on online network was established.

In this study, we constructed a simple and practical tool for predicting infectious complications after radical gastrectomy of gastric cancer, which can quantify the individual risk of postoperative infection and promote the implementation of early prevention and treatment.