2型糖尿病患者饮食依从性预测模型的建立及评价  

Construction and evaluation of a prediction model for diet compliance of patients with type 2 diabetes mellitus

作　　者：宋威 梁琳琅 冯聪 SONG Wei;LIANG Linlang;FENG Cong(Department of Endocrinology,General Hospital of Northern Theater Command of Chinese PLA,Shenyang 110016,China)

机构地区：[1]中国人民解放军北部战区总医院内分泌科,辽宁沈阳110016

出　　处：《现代医学》2025年第4期622-628,共7页Modern Medical Journal

基　　金：辽宁省科技计划项目(2023-MS-038)。

摘　　要：目的:基于2型糖尿病(T2DM)患者饮食依从性的影响因素,构建饮食低依从性的预测模型。方法:选取2019年6月至2023年6月在本院诊治的256例T2DM患者,出院后6个月进行饮食依从性的随访调查,并根据调查结果分成低依从组和非低依从组,筛选饮食低依从性的相关因素,并将其作为预测变量完成预测模型的构建,采用受试者工作特征(ROC)曲线下面积(AUC)、校正曲线和决策曲线对模型进行评价。结果:256例患者中78例(30.47%)饮食低依从;年龄≥60岁、学历为初中及以下、行动益处的感知得分低下、行动障碍的感知得分低下、低社会支持度、存在抑郁为T2DM患者饮食低依从的危险因素。AUC为0.832,模型预测饮食低依从的概率高度接近实际概率,拟合优度HL检验χ^(2)=7.643,P=0.469,阈值概率在4%~81%时模型净获益率较高。结论:根据年龄、学历、行动益处的感知、行动障碍的感知、社会支持度、是否存在抑郁构建的列线图可有效预测T2DM患者饮食低依从的概率。Objective:To construct a prediction model of low diet compliance based on the influential factors of diet compliance in patients with type 2 diabetes mellitus(T2DM).Methods:Totally 256 T2DM patients treated in our hospital from June 2019 to June 2023 were included,and a follow-up survey on diet compliance was conducted 6 months after discharge.Then they were divided into low compliance group and non-low compliance group according to the survey results.The relevant factors of low diet compliance were screened and used as predictive variables to construct the prediction model.The area under the receiver operating characteristic(ROC)curve(AUC),calibration curve and decision curve were applied to evaluate the model.Results:Among 256 patients,78(30.47%)had low diet compliance.Age≥60 years,education level of junior high school and below,low perception score of action benefit,low perception score of action disorder,low social support and depression were the risk factors for low diet compliance in T2DM patients.AUC was 0.832,the probability of low diet compliance predicted by the model was highly close to the actual probability,goodness of fit HL test showedχ^(2)=7.643,P=0.469,the net benefit rate of the model was higher at threshold probability of 4%-81%.Conclusion:The nomogram constructed by age,education,perception of action benefit,perception of action disorder,social support and presence of depression can effectively predict the probability of low diet compliance in patients with T2DM.

关 键 词：2型糖尿病 饮食依从性 预测模型 列线图 评价 

分 类 号：R473.5[医药卫生—护理学]