基于图论的艾滋病辨证论治辅助决策模型的研究

项目名称： 基于图论的艾滋病辨证论治辅助决策模型的研究

项目编号： No.81202858

项目类型： 青年科学基金项目

立项/批准年度： 2013

项目学科： 医学八处

项目作者： 赵玉凤

作者单位： 中国中医科学院中医临床基础医学研究所

项目金额： 23万元

中文摘要： 中医药治疗艾滋病已经有了多年的临床经验积累，形成了初步的辨证论治方法和相关方药，如何在借鉴前人经验的基础上继承创新，是提高临床治疗效果的关键。本研究将在近年来课题组已经收集到的1万多例中医药治疗艾滋病的临床诊疗病历数据基础上，利用图论方法建立艾滋病辨证论治辅助决策模型，为临床借鉴前人经验提高诊疗水平提供技术支持。研究将用能够有效利用部分临床有完整症状体征以及辨证论治方药的病历数据，作为已标记样本，与大量临床症状体征不够完整的病历数据，作为未标记样本，通过以患者为节点的空间流图学习，建立患者基本辨证与药物决策集合；通过以药物为节点的空间流图学习，建立具有完整语义关联的改善药物集合，进而揭示艾滋病症-证-治之间潜在关联关系的最优决策路径，形成能够体现辨证论治规律的辅助决策方法，为临床借鉴前人经验提高诊疗水平提供技术支持。

中文关键词： 艾滋病；辨证论治；数据挖掘；辅助决策；

英文摘要： The preliminary methods and related recipe of diagnosis and treatment had been acumulated in Traditional Chinese Medicine (TCM)to treat AIDS for many years. The heritance and innovation of the previous clinical experience is the key to improve AIDS clinical outcomes. More than ten thousand clinical records of the treatmeant based on syndrom differentiation, about the TCM treatment of AIDS had been collected by our team.Based on these records, the graph theory is used to verify the AIDS diagnosis and treatment decision support model, which is the technical base of diagnosis and treatment improment with the pervious experience. Some complete medical treatment prescription records of the treatmeant based on syndrom differentiation, which is named labeled-data, and a lot of incomplete medical records data of clinical signs and symptoms, which is named unlabeled-data, are combined to build the bsic decision set of syndromes and recipe by the spatial manifold of patient. Then, the improved recipe set with complete semantic raltion is found by the spatial manifold of drugs, which is ultilized to explore the optimal decision path of the potential relation among the symptom, the snydroms and the treatment. Finally, the decision suport model of treatmeant based on syndrom differentiation are constructed to improve the cli

英文关键词： AIDS；Sydrome Differentiation；Data Mining；Aid-decision；