面向医疗信息检索的本体构建和管理技术研究

发布时间：2018-11-21 07:12

【摘要】：由于医疗信息的特殊性导致对于其的检索效果不尽如人意,本体的出现使得局面有所改善。在构建本体的过程中要尽量做到对现有本体的复用,以达到全面、正确、快速构建本体的目的,本体间的异构性可以通过本体映射算法来解决。虽然已经存在了很多本体映射算法的研究成果,但是大多集中在相似度度量原理的设计上,对于其它方面关注甚少。本文针对现有本体映射算法在语义描述全面性、映射过程安全性和概念识别增量性方面存在的不足,对本体映射算法进行了改进,提出了具有隐私保护的模糊本体增量式映射算法,利用病历数据进行了分类实验,给出了结果说明,并且分析了此算法的优缺点。本文介绍了医疗信息检索的研究背景、目的意义和国内外研究现状,从理论和应用两个方面说明了制约循证医学发展的因素,强调了本文研究内容的必要性;通过对医疗信息组织形式的介绍,突出了本体在描述医疗信息方面的优势;通过对本体匹配原理的分析,得知了现存相似性计算的局限性,确定了具体的研究思路;经过分析目前现有实验资源和数据支持,估计了研究过程中可能遇到的难题,在一定程度上规避了风险,增强了自信心。为了改进目前本体相似性度量中存在的不足,在现有四元结构本体中加入模糊概念,形成了新的五维本体,增强了本体的语言描述能力;在映射过程中引入混淆思想和适合的相似度度量方法,在保证映射正确的前提下提高了映射的安全性;在概念识别过程中,加入增量识别原理,完成了对本体中未定义概念的识别;结合以上三点,提出了具有隐私保护的模糊本体增量式映射算法,并详述了其工作原理和实现过程。本文通过对医学主题词表、统一医学语言系统和国际疾病分类编码与术语表第十版的相关知识体系的复用,构建了“先天性心脏缺损”疾病病历本体,完整再现了本体的映射建立过程,完成了基于电子病历文档的概念识别和语义网络的构建,丰富了医疗领域本体构成。并且本文进行了原型系统的开发,以心内科病历数据作为研究对象进行实验,描述了分类过程,对实验结果进行了分析,可以看出本文所提算法的有效性和可行性得到了证明。
[Abstract]:Because of the particularity of medical information, the retrieval effect of medical information is not satisfactory, and the appearance of ontology improves the situation. In the process of building ontology, we should reuse existing ontology as far as possible to achieve the goal of building ontology comprehensively, correctly and quickly. The heterogeneity between ontologies can be solved by ontology mapping algorithm. Although there have been many researches on ontology mapping algorithms, most of them are focused on the design of similarity measurement principle, and little attention has been paid to other aspects. Aiming at the shortcomings of the existing ontology mapping algorithms in semantic description, mapping process security and concept recognition increment, the ontology mapping algorithm is improved in this paper. A fuzzy ontology incremental mapping algorithm with privacy protection is proposed. The classification experiment is carried out using medical record data. The results are illustrated and the advantages and disadvantages of the algorithm are analyzed. This paper introduces the research background of medical information retrieval, its significance and research status at home and abroad, explains the factors restricting the development of evidence-based medicine from two aspects of theory and application, and emphasizes the necessity of the research content of this paper. Through the introduction of the organization form of medical information, the advantages of ontology in describing medical information are highlighted, through the analysis of ontology matching principle, the limitations of existing similarity calculation are known, and the specific research ideas are determined. Based on the analysis of the current experimental resources and data support, this paper estimates the difficulties that may be encountered in the research process, avoids the risk to a certain extent, and strengthens the self-confidence. In order to improve the shortcomings of ontology similarity measurement, a new five-dimensional ontology is formed by adding fuzzy concept to the existing four-element ontology, and the language description ability of ontology is enhanced. The confusion idea and the suitable similarity measure method are introduced in the mapping process, which improves the security of the mapping under the premise of ensuring the mapping is correct. In the process of concept recognition, incremental recognition principle is added to realize the recognition of undefined concepts in ontology. Combined with the above three points, a fuzzy ontology incremental mapping algorithm with privacy protection is proposed, and its working principle and implementation process are described in detail. Based on the reuse of the related knowledge system of the medical subject vocabulary, the unified medical language system and the codes of the International Classification of Diseases and the tenth edition of the glossary, this paper constructs the Noumenon of the medical record of "congenital heart defect". The mapping process of ontology is completely reproduced, the concept recognition based on electronic medical record document and the construction of semantic network are completed, which enriches the constitution of medical domain ontology. In addition, the prototype system is developed in this paper. Taking the medical records of cardiology as the research object, the classification process is described, and the experimental results are analyzed. The validity and feasibility of the proposed algorithm are proved.
【学位授予单位】：中北大学
【学位级别】：硕士
【学位授予年份】：2017
【分类号】：TP391.3

【相似文献】