北京生物医学工程2025，Vol.44Issue(3)：284-289,6.DOI:10.3969/j.issn.1002-3208.2025.03.010
基于SMOTE算法构建糖尿病足溃疡患者风险预测模型的初步探索
Preliminary exploration of constructing risk prediction model of diabetic foot ulcer based on SMOTE algorithm
摘要
Abstract
Objective To explore the independent risk factors of diabetic foot ulcers(DFU)in diabetic patients,and establish a warning model based on SMOTE algorithm.Methods The clinical data of 272 diabetic patients admitted to Mianyang Central Hospital from February 2017 to February 2021 were retrospectively analyzed.According to whether the patients had DFU,the patients were divided into DFU group and non-DFU group.Multivariate Logistic regression analysis was used to screen the independent risk factors of diabetic DFU and construct the prediction model X1.At the same time,improve the data set based on SMOTE algorithm,build the early warning model X2,and compare and verify the prediction efficiency of the model.The internal verification method adopts 5-fold cross-validation.Results Hypertension,peripheral neuropathy,C-reactive protein(CRP)≥8 mg/L,glycosylated hemoglobin(Type A1C),glycosylated hemoglobin(CRP)≥8 mg/L,HbA1c≥6%and Fibrinogen(FIB)≥4 g/L were independent risk factors for DFU in diabetic patients(P＜0.05).The area under ROC curve of early warning model X1 based on the above risk factors was 0.888(95%CI:0.826-0.951),and the area under ROC curve of early warning model X2 based on the SMOTE algorithm improved data set was 0.916(95%CI:0.891-0.941),both of which had higher prediction accuracy,but the prediction effect of the latter was slightly higher than that of the former.The accuracy rate of the model was 77.2%.Conclusions Hypertension or elevated blood pressure,peripheral neuropathy,high CRP,HbA1c and FIB levels are independent risk factors for diabetic foot ulcers,which should be paid attention to in clinical treatment.The risk early warning model based on SMOTE over-sampling algorithm has high predictive efficacy,clinical can make corresponding prevention and cure measures according to the above mentioned high risk group of foot ulcer.关键词
SMOTE算法/糖尿病足溃疡/Logistic分析/风险预测模型Key words
SMOTE algorithm/diabetic foot ulcer/logistic analysis/risk prediction model分类
医药卫生引用本文复制引用
郭福蝶,李希西,李叶萍,罗杨,张红梅,白迎春..基于SMOTE算法构建糖尿病足溃疡患者风险预测模型的初步探索[J].北京生物医学工程,2025,44(3):284-289,6.基金项目
四川省医学会伤口疾病(泰阁)专项课题(2021TG60)、绵阳市中心医院科研课题(2019YJ09)、绵阳市卫生计生委科研课题(201749)资助 （泰阁）
