基于SMOTE的髋部骨折病人术后便秘预测模型的构建

Construction of constipation risk prediction model for hip fracture patients after surgery based on SMOTE

目的:基于人工少数类过采样法(SMOTE)构建髋部骨折病人术后便秘预测模型。方法:选取2020年1月—2021年12月我院收治的130例髋部骨折术后病人作为研究对象,根据病人术后30 d内是否发生便秘分为便秘组(41例)和非便秘组(89例)。收集两组病人资料,采用单因素分析和二元Logistic回归分析方法筛选病人术后发生便秘的独立危险因素,并建立Logistic回归模型(P_(1)),基于SMOTE改进数据集,构建基于SMOTE的Logistic预测模型(P_(2)),验证模型的预测效能。结果:年龄(X_(1))、糖化血红蛋白(X_(2))、是否合并使用阿片类镇痛药物(X_(3))、术后卧床时间(X_(4))和术后首次进食时间(X_(5))是髋部骨折病人术后发生便秘的独立影响因素(P<0.05),P_(2)=1/[1+e-(-35.244+0.163X+1.469X+1.562X+0.313X+0.340X)],P2的决定系数为0.904,受试者工作特征曲线下面积为0.936,均高于P1(决定系数为0.878,受试者工作特征曲线下面积为0.752)。结论:基于SMOTE构建的髋部骨折病人术后便秘预测模型对髋部骨折病人术后便秘发生风险具有较高预测效能。

Objective:To construct constipation risk prediction model for hip fracture patients after surgery based on synthetic minority over-sampling technique(SMOTE).Methods:A total of 130 hip fracture patients after surgery admitted to our hospital from January 2020 to December 2021 were selected as research subjects.According to whether patients had constipation within 30 days after surgery or not,they were divided into constipation group(41 cases)and non-constipation group(89 cases).The data of the two groups of patients were collected.The independent risk factors of constipation after surgery were screened by univariate analysis and binary Logistic regression analysis,and a Logistic regression model(P_(1))was constructed.The data set was improved based on SMOTE,and a Logistic prediction model(P_(2))based on SMOTE was constructed.Prediction efficiency of the model were tested.Results:Age(X_(1)),glycosylated hemoglobin level(X_(2)),whether combined use opioid analgesics or not(X_(3)),time in bed after operation(X_(4))and first feeding time after operation(X_(5))were independent influence factors of constipation for hip fracture patients after surgery(P<0.05),P_(2)=1/[1+e-(-35.244+0.163X+1.469X+1.562X+0.313X+0.340X)],the coefficient of determination of P2 was 0.904,and the area under receiver operator characteristic was 0.936,both higher than P1(coefficient of determination was 0.878,and the area under receiver operator characteristic was 0.752).Conclusions:The constipation risk prediction model for hip fracture patients after surgery based on SMOTE has higher prediction efficiency on constipation risk for hip fracture patients after surgery.