摘要
目的 分析糖尿病肾病的强影响因素,并对多种预测模型进行评价,选择预测效率最佳的模型,为构建糖尿病并发症预测系统、实现糖尿病肾病早期预防提供参考。方法 选取单因素分析有统计学意义的指标分别构建Logistic回归模型、神经网络模型和基于Logistic回归分析的神经网络模型。结果 3种预测模型对训练组的正确率分别为83.4%、81.3%、83.7%,神经网络模型和基于Logistic回归分析的神经网络模型对于测试组的正确率为85.5%、86.0%。微量尿蛋白、血尿素和血肌酐3个指标在3种预测模型中均有意义。结论 神经网络模型的预测效果优于Logistic回归模型,基于Logistic回归分析的神经网络模型可以在保证预测效果相当的情况下,减少输入指标数量和运算时间,有效提高预测效率。微量尿蛋白、血尿素和血肌酐等生化指标可以为临床糖尿病肾病的预防与诊断提供参考。
Objective To analyze the strong influencing factors of diabetic nephropathy and evaluate multiple prediction models,in order to select the best prediction efficiency model for constructing a prediction system for diabetic complications and achieving early prevention of diabetic nephropathy.Methods Select the statistical significance indicators for univariate factor analysis to respectively construct logistic regression models,neural network model and neural network models based on logistic regression analysis.Results The accuracy of three prediction models for the training group were respectively 83.4%,81.3%,and 83.7% and for the test group the accuracy of the neural network model and the neural network model based on logistic regression analysis were 85.5% and 86.0%.Microalbuminuria,blood urea nitrogen and blood creatinine,these three indicators w ere significant in all three prediction models.Conclusion The prediction effect of the neural network model is better than the logistic regression model,and the neural network model based on logistic regression analysis could reduce the number of input indicators and operation time to effectively improve the prediction efficiency while ensuring comparable prediction effect.Biochemical indicators like microalbuminuria,blood urea nitrogen and blood creatinine could provide preference for the prevention and diagnosis of clinical diabetic nephropathy.
作者
国航
任敬
侯晓丽
崔倩倩
王辛哲
孔杨
GUO Hang;REN Jing;HOU Xiaoli;CUI Qianqian;WANG Xinzhe;KONG Yang(School of Public Health and Management of Binzhou Medical University,Yantai City,Shandong Province,China264003)
出处
《卫生职业教育》
2023年第1期155-159,共5页
HEALTH VOCATIONAL EDUCATION
基金
教育部人文社会科学研究青年基金项目(19YJCZH072)
国家级大学生创新训练项目(202210440030)。
关键词
神经网络
糖尿病肾病
预测模型
Neural network
Diabetic nephropathy
Prediction models
