摘要
目的糖尿病是威胁中老年人健康的一种常见的慢性非传染性疾病,为实现对糖尿病的早预防、早发现、早治疗,本研究基于实证调查数据,构建了一个中老年人罹患糖尿病的风险预测模型。方法在浙江省宁波市海曙区集士港镇共计调查了2567例中老年人的糖尿病患病情况及基本人口学信息、生活方式、既往病史、社会心理变量等数据,使用logistics逻辑回归方法进行风险预测模型的构建。结果模型在训练集的预测准确率为0.65,Kappa值为0.13,灵敏度为0.66,特异性为0.58,精确度为0.91,曲线下面积为0.67,在测试集的预测准确率为0.64,Kappa值为0.12,灵敏度为0.66,特异性为0.54,精确度为0.89,曲线下面积为0.64。社会人口学变量如年龄、中心性肥胖,饮食习惯和生活习惯如酒精使用等是糖尿病显著的风险因素,模型预测的准确度达到63.94%,预测为阳性的患者有89.26%的概率确为糖尿病患者。结论模型的预测准确度良好,年龄、中心性肥胖、饮食习惯和酒精使用对糖尿病的发生有重要预测作用。
Objective Diabetes is a common chronic condition which impacts the health of the middle-aged and elderly.To prevent,recognize,and treat the disease,a predictive model for the population was built based on an empirical investigation.Methods A total of 2567 middle-aged and elderly were investigated.The diagnosis of diabetes,as well as the demographic characteristics,lifestyles,anamnesis,and psychosocial variables were collected.Logistic regression analyses were performed to build the model.Results The prediction accuracy of the model in the training set is 0.65,Kappa value is 0.13,sensitivity is 0.66,specificity is 0.58,accuracy is 0.91,and area under the curve is 0.67.The prediction accuracy in the test set is 0.64,Kappa value is 0.12,sensitivity is 0.66,specificity is 0.54,accuracy is 0.89,and area under the curve is0.64.Demographic characteristics,such as age,central obesity,dietary habits,lifestyle,and alcohol use,were significantrisk factors for diabetes.The prediction accuracy of the model was 63.9%,while 89.3%of those with positive scores were actual patients with diabetes.Conclusion Themodel has an acceptable accuracy of prediction;age,central obesity,dietary habits,and alcohol use are important predictive factors for diabetes.
作者
邬锡波
邓旭
韩红霞
王佳宇
薛将
陈树林
WU Xi-bo;DENG Xu;HAN Hong-xia;WANG Jia-yu;XUE Jiang;CHEN Shu-lin(Ningbo Haishu No.3 Hospital,Zhejiang 315191,China)
出处
《慢性病学杂志》
2022年第9期1335-1341,共7页
Chronic Pathematology Journal
