糖尿病健康数据分析方法及应用

发布时间：2018-03-14 13:46

本文选题：糖尿病　切入点：健康数据分析　出处：《哈尔滨工业大学》2017年硕士论文　论文类型：学位论文

【摘要】：随着大众健康意识的日益提高,普通公民对糖尿病健康服务也提出了新的需求。人们希望能尽早预知糖尿病风险,足不出户地掌握自身病情状态。为满足这一需求,智能健康产业应运而生。当前,互联网时代积累了大量的糖尿病健康数据,包括诊断记录、病历信息、电子档案等;各种各样的健康监测设备层出不穷,实现了个人日常健康信息的随时收集和存储。如何充分利用上述数据,为人们提供便利的糖尿病自主评估服务,进而实现降低我国糖尿病发病率的目的,成为当前亟待解决的问题。应用计算机技术对糖尿病健康数据进行分析,是实施健康预测和辅助诊断的有效解决方案。本文结合现有的糖尿病医学评估工具的不足,对糖尿病健康数据分析方法展开研究,并进行对应的系统开发,以提供糖尿病相关的健康咨询服务。本研究主要涉及以下几个方面:首先,为了解决糖尿病的风险识别和预测问题,进行糖尿病风险计算方法的研究。在抽象化糖尿病输入信息并量化风险参数的基础上,挖掘实际健康数据中的规律,建立一个基于支持向量机(SVM)的糖尿病风险计算模型。该模型处理用户的输入记录,计算出用户患糖尿病的风险,以实现糖尿病早识别、早预防和早治疗的目标。其次,为了合理利用糖尿病的遗传特征,确保糖尿病风险预测的准确性,建立一种糖尿病遗传因素提取机制。该机制结合相关的医学知识,通过追溯糖尿病家族史绘制遗传关系图。并提出相应的遗传特征提取算法,用于充分提取用户的先天性疾病信息。将该机制用于糖尿病风险计算模型,将有效地提高模型的综合性能。再次,为了实现对糖尿病动态疾病信息的预测,进行动态血糖预测方法的研究。提出动态血糖预测模型,将动态血糖数据进行提取和表达,并基于深度信念网(DBN)探索现有的血糖时间序列,以预知未来一段时间内的血糖。该模型可以帮助用户动态掌握和预测血糖水平。最后,在上述工作的基础上,设计并实现一个糖尿病辅助评估系统,从架构、数据库、业务功能、用户接口等方面完成算法应用化。
[Abstract]:With the increasing awareness of public health, ordinary citizens have put forward a new demand for diabetes health services. People hope to be able to predict the risk of diabetes as soon as possible, and master their state of illness without leaving home. The intelligent health industry came into being. At present, the Internet era has accumulated a large amount of diabetes health data, including diagnostic records, medical records, electronic files, etc. How to make full use of the above data, and how to provide people with convenient diabetes assessment services, and then achieve the goal of reducing the incidence of diabetes in China. The application of computer technology to the analysis of diabetes health data is an effective solution for the implementation of health prediction and auxiliary diagnosis. Research on diabetes health data analysis method, and corresponding system development to provide diabetes related health counseling services. This study mainly involves the following aspects: first, In order to solve the problem of risk identification and prediction of diabetes mellitus, the risk calculation method of diabetes mellitus is studied. On the basis of abstracting the input information of diabetes mellitus and quantifying the risk parameters, the rules of actual health data are excavated. A diabetes risk calculation model based on support vector machine (SVM) is established. The model processes the user's input record and calculates the user's risk of developing diabetes, so as to achieve the goal of early recognition, early prevention and early treatment of diabetes. In order to make rational use of the genetic characteristics of diabetes mellitus and ensure the accuracy of diabetes risk prediction, a mechanism for extracting genetic factors from diabetes mellitus was established. By tracing back the family history of diabetes, the genetic relationship map was drawn, and the corresponding genetic feature extraction algorithm was put forward, which was used to fully extract the information of the user's congenital disease, and the mechanism was applied to the diabetes risk calculation model. It will improve the comprehensive performance of the model effectively. Thirdly, in order to predict the dynamic disease information of diabetes mellitus, the dynamic blood glucose prediction method is studied. A dynamic blood glucose prediction model is proposed to extract and express the dynamic blood sugar data. And based on DBN (depth belief Network), we explore the existing blood glucose time series to predict the blood sugar in the future. This model can help users to grasp and predict the blood sugar level dynamically. Finally, based on the above work, A diabetes aided evaluation system is designed and implemented. The algorithm is applied from the aspects of architecture, database, business function, user interface and so on.
【学位授予单位】：哈尔滨工业大学
【学位级别】：硕士
【学位授予年份】：2017
【分类号】：R587.1;TP311.13

【相似文献】