基于数据挖掘技术的生物反馈治疗辅助系统的设计与实现

发布时间：2018-03-09 05:11

本文选题：医学数据挖掘　切入点：BP神经网络算法　出处：《中山大学》2012年硕士论文　论文类型：学位论文

【摘要】：随着生物医学工程的迅猛发展，测量仪器技术的提高，大量医疗数据被精确地记录下来，从而导致医疗数据资料爆炸性增长，因此数据挖掘技术被广泛应用于医学领域以发现海量数据中潜在的知识。在本院与中山大学医学院合作的交叉学科项目中，医学院提出了采用新兴的生物反馈疗法对高血压前期进行干预作用，在治疗中医生通过引导患者调节心率变异性（HRV）以达到调节血压的功效。这一治疗方法已经积累了（并且将持续积累）大量治疗方案、生物体征数据等，本文的研究动机就是将数据挖掘技术应用在治疗数据上，通过这些数据预测患者治疗的有效性以及治疗后的HRV值，为医生在治疗过程中的决策提供指导依据。本文在学习了生物反馈领域知识的基础上，提取出两个挖掘任务并建立了分类和预测模型，主要流程如下：1）根据领域知识进行特征选择，进行数据预处理后建立治疗有效性的分类预测模型以此提高医生治疗的针对性，并且对比了C4.5算法和随机森林算法，实验显示随机森林算法模型的准确率高于C4.5；2）以患者治疗前，治疗中的各体征值建立HRV值的回归预测模型来帮助医生更准确地找到HRV目标值，并且对比了BP神经网络算法和多元回归算法，实验显示BP神经网络算法误差较小；3）最后设计与实现了生物反馈治疗辅助系统，，以更好的人机交互和操作流程可视化方式将效果较好的随机森林算法和BP神经网络算法模型应用在该系统中。本文依照实证研究的方法，在26个患者治疗过程实例中收集医生预测的结果，并对比本文系统的预测结果。实验证明本文分类和预测模型的准确率都高于医生的经验预测结果，因此预测结果在医生设计生物反馈治疗方案过程中起到了指导作用，有一定的临床意义。同时本文实现的治疗反馈辅助系统将数据挖掘知识应用在实际中，集成了治疗过程中情绪问卷调查、血压对比等功能，优化了医生的治疗流程，提高了研究疗效的效率。
[Abstract]:With the rapid development of Biomedical Engineering, instrumentation technology, a large number of medical data are accurately recorded, resulting in the explosive growth of medical data, so data mining technology has been widely applied in the field of medicine in order to find out the potential massive data knowledge. In this interdisciplinary project in cooperation with the Institute of Zhongshan University School of Medicine School of medicine, put forward the emerging bio feedback therapy intervention effect on hypertension in the early treatment of Chinese students by guiding the patients to regulate the heart rate variability (HRV) in order to regulate the blood pressure effect. This treatment method has been accumulated (and will continue to accumulate a large number of) treatment, biometric data, study motivation this paper is the application of data mining technology in the treatment of the data, through these data to predict patient effectiveness of treatment and after treatment HRV value for doctors Provide guidance for decision making in the course of treatment.
In this paper, learning the basic knowledge in the field of biological feedback, extract two mining tasks and established a classification and prediction model, the main process is as follows: 1) feature selection based on domain knowledge, data preprocessing is carried out after the establishment of a prediction model for the effectiveness of this to improve the relevance of medical treatment, and compared the C4.5 algorithm and random forest algorithm, experiments show that the accuracy of the random forest algorithm model is higher than that of C4.5; 2) in patients before treatment, the symptoms in the treatment of value regression HRV value prediction model to help doctors more accurately find the target value of HRV, and compared with the BP neural network algorithm and regression algorithm, experiments show BP neural network algorithm has smaller error; 3) the design and Realization of the biofeedback assisted system, in order to better human-computer interaction and operation flow visualization will better effect with The computer forest algorithm and the BP neural network algorithm model are applied to the system.
According to the method of empirical research, in 26 of patients in the process instance to collect the doctor predicted results, and compared the system prediction results. The experiment results show that the prediction results of the doctor is higher than the accuracy of this classification and prediction model of the experience, so the prediction result has played a guiding role in the design of the doctor biofeedback treatment process, has a certain clinical significance. The treatment and the auxiliary feedback system will be the knowledge of data mining applications in integrated treatment of emotion questionnaire survey, comparison of blood pressure and other functions, optimizing the medical treatment process, improve the efficiency of treatment.

【学位授予单位】：中山大学
【学位级别】：硕士
【学位授予年份】：2012
【分类号】：R318.6;TP311.13

【参考文献】