卵巢癌术后发生胃肠功能紊乱风险预警模型的构建与对比分析  

Construction and comparative analysis of a risk warning model for gastrointestinal dysfunction in postoperative patients with ovarian cancer

作　　者：卢淑兰[1] 郑建敏[1] 余丽[1] 罗红莲[1] Lu Shulan;Zheng Jianmin;Yu Li;Luo Honglian(Department of Gynecology,First Affiliated Hospital of Guangxi Medical University,Nanning Guangxi 530021,P.R.China)

机构地区：[1]广西医科大学第一附属医院妇科一病区,广西南宁530021

出　　处：《中国计划生育和妇产科》2025年第1期92-95,101,共5页Chinese Journal of Family Planning & Gynecotokology

基　　金：广西壮族自治区卫生健康委员会自筹经费科研课题(项目编号:Z20210772)。

摘　　要：目的基于3种统计学方法构建卵巢癌患者术后发生胃肠功能紊乱风险预警模型,并比较其预测效能。方法回顾性选取2021年9月至2022年8月在广西医科大学第一附属医院妇科病区行卵巢癌手术的240例患者为入组对象,根据术后是否发生胃肠功能紊乱分为病例组和对照组。同时按7∶3的比例将240例患者分为训练集和测试集,训练集用于构建模型,测试集用于验证模型。收集患者临床资料,通过对相关危险因素分析获得P<0.05的独立预测因素,分别运用Logistic回归、随机森林及决策树模型建立卵巢癌患者术后发生胃肠功能紊乱风险预警模型,并分析其预测价值。结果两组在术前营养不良、麻醉时间、盆腔淋巴结清扫、尿pH、血钠、血钾、胃泌素方面比较,差异有统计学意义(P<0.05)。运用测试集对3种模型进行效能评价,得出Logistic回归模型预测的特异度为0.839,敏感度为0.878,AUC为0.931;随机森林模型预测的特异度为0.903,敏感度为0.878,AUC为0.945;决策树模型预测的特异度为0.756,敏感度为0.806,AUC为0.793。经比较,随机森林模型的风险预警效果更为理想,其在卵巢癌患者术后胃肠功能紊乱风险识别中的诊断价值高于Logistic回归与决策树模型。结论随机森林模型在卵巢癌患者术后发生胃肠功能紊乱风险预警和评估中具有较高的应用价值,其预测效能优于Logistic回归模型与决策树模型。Objective To construct a risk warning model for gastrointestinal dysfunction in patients after ovarian cancer surgery based on 3 statistical methods and analyze their predictive efficacy.Methods A retrospective study was conducted to select 240 patients with ovarian cancer who underwent surgery in the gynecological ward of First Affiliated Hospital of Guangxi Medical University from September 2021 to August 2022.According to whether gastrointestinal dysfunction occurred after operation,the patients were divided into case group and control group.At the same time,240 patients were divided into training set and testing set according to the ratio of 7∶3.The training set was used to build the model,and the testing set was used to validate the model.Clinical data of patients were collected,and the independent predictive factors with P<0.05 were obtained through the analysis of relevant risk factors.The risk warning model of gastrointestinal dysfunction in patients with ovarian cancer after surgery was established by using Logistic regression,random forest and decision tree models respectively,and their predictive value was analyzed.Results There was a statistically significant difference between the two groups in preoperative malnutrition,anesthesia time,pelvic lymph node dissection,urine pH,blood sodium,blood potassium,and gastrin(P<0.05).The testing set was used to evaluate the efficacy of the 3 models,and the specificity,sensitivity and AUC of the Logistic regression model were 0.839,0.878 and 0.931,respectively.The specificity,sensitivity and AUC of the random forest model were 0.903,0.878 and 0.945,respectively.The specificity,sensitivity and AUC of the decision tree model were 0.756,0.806 and 0.793,respectively.By comparison,the random forest model had a better risk warning effect,and its diagnostic value in the risk identification of postoperative gastrointestinal dysfunction in patients with ovarian cancer was higher than that of Logistic regression and decision tree model.Conclusion The random forest model h

关 键 词：卵巢癌 胃肠功能紊乱 风险预警 预测模型 

分 类 号：R737.31[医药卫生—肿瘤]