面向高血压的慢性病管理辅助决策系统的研究

发布时间：2018-04-24 23:16

本文选题：慢性病管理 + 高血压　；参考：《电子科技大学》2017年硕士论文

【摘要】：高血压患者血压自我管理意识淡薄,血压得不到有效控制将引起一系列常见并发症。医患沟通缺乏会导致医生不能及时掌握患者病情发展,无法开展患者的个性化精准医疗指导活动。如何使用信息化手段帮助高血压患者提高血压自我管理水平,形成良好的医患互动模式已成为一个亟待解决的重要问题。基于Android平台、Visual Studio 2013工具和MySQL数据库设计实现面向高血压的慢性病管理辅助决策系统,服务端采用WCF架构。系统分为患者端和医生端,患者端实现血压心率等体征数据采集与监控、行为监控、高血压慢性病风险评估和预警等功能;医生端实现辅助医生对患者进行健康指导功能和制定医嘱。论文研究内容主要包括高血压慢性病风险因素提取、高血压慢性病分类诊断、高血压慢性病风险评估三个模块。具体研究工作如下:1.引入互信息、遗传算法(Genetic Algorithm,GA)和朴素贝叶斯(Bayes Na?ve,BN)算法对高血压慢性病风险因素进行特征提取,互信息是遗传算法特征选择的前期准备,NB算法为GA算法中特征子集的评价函数。与最优优先搜索(Best First Search,BFS)算法、序列浮动前向选择(Sequential Floating Forward Selection,SFFS)进行比较。通过特征选择分类精度分别为87.50%,83.92%,85.71%。2.引入基于加权投票表决的分类器融合算法,基于支持向量机(Support Vector Machine,SVM)、K最近邻(k-NearestNeighbor,KNN)、朴素贝叶斯(Bayes Na?ve,BN)和反向传播神经网络(Back-Propagation Neural Networks,BPNN)四个单分类算法进行分类器训练,通过分类器投票加权和分类排名两种策略改进算法,基于改进的加权投票表决算法预测准确率提高5%以上。3.引入最小二乘法构建1-1模型,计算单风险因素与高血压慢性病之间的关系,引用朴素贝叶斯算法构建n-1模型,评估多因素与高血压慢性病之间的关系。从中提取出与高血压慢性病有强相关度的因素来进行疾病诊断和风险评估,辅助医生进行高血压慢性病的诊断和治疗。
[Abstract]:The self-management consciousness of blood pressure in hypertensive patients is weak and the blood pressure can not be effectively controlled will cause a series of common complications. The lack of communication between doctors and patients will lead to the doctor can not grasp the patient's condition in time, and can not carry out the patient's individualized precise medical guidance. How to use informational means to help the hypertensive patients improve their blood pressure self Management level, forming a good medical and patient interaction model has become an important problem to be solved urgently. Based on the Android platform, Visual Studio 2013 tools and MySQL database design and implement the chronic disease management assistant decision-making system for hypertension. The server adopts the WCF architecture. The system is divided into the patient end and the doctor side, the patient end realizes the blood pressure heart. Rate and other functions, such as data collection and monitoring, behavior monitoring, risk assessment and early warning of hypertension chronic diseases, and doctors to assist doctors to carry out health guidance functions and make medical orders. The main contents of the thesis include the extraction of risk factors of chronic hypertension, the classification diagnosis of high blood pressure slow disease, and the risk assessment of hypertension chronic disease. Three modules are estimated. The specific research work is as follows: 1. introduction of mutual information, genetic algorithm (Genetic Algorithm, GA) and simple Bias (Bayes Na VE, BN) algorithm for characteristic extraction of the risk factors of hypertension chronic disease, mutual information is the pre preparation of the genetic algorithm feature selection, NB algorithm is the evaluation function of the feature subset in the GA algorithm. The first search (Best First Search, BFS) algorithm, sequence floating forward selection (Sequential Floating Forward Selection, SFFS) are compared. The classification accuracy is 87.50%, 83.92%, and 85.71%.2. introduces the classifier fusion algorithm based on the weighted voting, and the nearest neighbor is based on the support vector machine (Support Vector). -NearestNeighbor, KNN), four single classification algorithms of simple Bias (Bayes Na? VE, BN) and back propagation neural network (Back-Propagation Neural Networks, BPNN) are trained for classifier, and two strategies are improved by classifier voting weighting and classification ranking, and the accuracy rate is increased by more than 5%.3 based on the improved weighted voting algorithm. The 1-1 model was constructed with the least square method to calculate the relationship between the single risk factors and the hypertensive chronic disease. The naive Bayes algorithm was used to construct the N-1 model, and the relationship between the multiple factors and the hypertension chronic disease was evaluated. The diagnosis and treatment of high blood pressure chronic disease.

【学位授予单位】：电子科技大学
【学位级别】：硕士
【学位授予年份】：2017
【分类号】：R544.1;TP311.52

【参考文献】