基于机器学习的糖尿病并发症预测模型研究进展

Research progress of prediction model of the diabetic complications based on machine learning

针对2型糖尿病(T2DM)并发症的诊断预测问题,传统检测方法主要通过血液和尿液检查来预测,这些方法既耗时又不能进行早期预测。目前,由于糖尿病发病率升高以及医疗数据的大幅增加,机器学习算法迅速发展为检测及诊断糖尿病的有效方法。用机器学习算法分析临床指标,探究2型糖尿病并发症的影响因素,构建并发症预测模型,可以很好地实现糖尿病并发症预防。通过介绍近年来预测及诊断糖尿病并发症的机器学习算法,对贝叶斯网络、随机森林、支持向量机以及反向传播(BP)神经网络4种机器学习优化算法分别进行概括讨论,以期提高对糖尿病及其预测模型的理解并降低疾病风险。

For the problems of diagnosis and prediction on complications of type 2 diabetes mellitus(T2DM),the conventional testing methods mainly predicted that through blood and urine tests,which are time-consuming and which doesn’t perform early prediction.At present,machine learning algorithms have rapidly developed into effective methods of detecting and diagnosing diabetes mellitus due to the increased incidence of diabetes mellitus and the large increase in medical data.The machine learning algorithm was used to analyze the clinical indicators,and to explore the influencing factors of the complications of type 2 diabetes mellitus,and to construct a prediction model of complication,which can commendably realize the prevention for complications of diabetes mellitus.This study comprehensively discussed 4 kinds of optimal algorithms of machine learning included Bayesian network,random forest,support vector machine and back propagation(BP)neural network,respectively,through introduced the algorithms of machine learning of predicting and diagnosing complications of diabetes mellitus in recent years,so as to improve the understanding for diabetes mellitus and its prediction model,and to reduce the disease risk.