DEVELOPMENT OF A MONITORING SYSTEM OF CLINICAL INFORMATION,INTEGRATED WITH HOSPITAL INFORMATION SYSTEM

(1)内科领域症状-疾病知识库的细化和医学术语匹配算法的开发我们开发了一个专门的知识库（KB）编辑器，用于细化内科领域知识库，该知识库是前一年开发的。使用此知识库编辑器，可以分析症状特征的表示，并将知识库中使用的症状分类为类似的组。提出了医学术语语义匹配算法，实现了医学术语的分组。该算法基于将一个复杂的术语分解为几个语义上基本的术语的技术。为了应用该算法，还利用本院信息系统的主数据库和医学电子词典开发了一套基本的医学术语集。(2)从症状产生疾病假设系统的实验开发在上述知识库中，我们开发了一个从症状产生疾病假设的原型系统。通过这一系统，揭示并证实了一种从临床并发症或临床表现推断可能疾病的过程。(3)利用医院信息系统中存储的临床数据进行系统的实验开发。一个原型系统，对医院信息系统中存储的临床事件进行监测，并对临床检查结果进行报警报告。该系统由三个子系统组成：第一部分是规则库系统，当某一临床结果在某一时刻满足一定条件时，规则库系统将启动预先定义的动作；第二部分是对规则库进行监控的系统；第三部分是从医院信息系统中提取必要数据目录的子系统。该监测系统的一个特点是与时间推理相结合，能够比较来自不同时间点的两个临床数据。(4)基于HL7的监护系统与医院信息系统之间的消息通信系统我们尝试实现美国使用的HL7消息通信协议的日本修改版本，这是日本的第一次实际试验。上面描述的推理引擎工作在UNIX工作站上，进行推理所需的信息使用这种修改后的HL7消息通信协议从医院信息系统传输。少

(1) Refinement of symptom-disease knowledge-base in internal medicine domain, and a development of algorithm for matching medical termsWe developed a specific knowledge-base (KB) editor for the refinement of the KB in internal medicine, the KB which was developed in the previous year. Using this KB editor, representation of symptom characteristics are analyzed, and that of symptoms used in the KB are categorized into similar groups. Algorithm for matching semantics of medical terms was developed to make groups of the terms. This algorithm is based a technique of breaking a complicated term into several semantically primitive terms. To apply this algorithm, a basic set of medical terms is also developed using a master database of our hospital information system and Medical Electronic Dictionary.(2) A experimental development of a system which generates disease hypotheses from symptomsUsin the above KB,we developed a prototype system which generates disease hypotheses from symptoms. Thro … More ugh this system, a process, which infers probable diseases from clinical complications or from clinical manifestations, was revealed and confirmed.(3) A experimental development of a system which utilize clinical data stored in hospital information systemA prototype system, which monitors clinical events and tells alarming reports about results of clinical examinations stored in hospital information system. This system consists of three sub-components : the first is a rule-base system which is designed to activate a previously defined action when a certain condition is satisfied about some clinical results at a certain time, the second is a monitoring system which watches the rule-base and the third is a sub-system which extracts necessary data directory from hospital information system. A feature of this monitoring system is integrated with temporal reasoning which is able to compare two two clinical data derived from different time points.(4) HL7-based message communication system between monitoring system and hospital information systemWe tried to implement a Japanese modified version of HL7 message communication protocol which is used in U.S.This implementation is a first prctical trial in Japan. The inferring engine described above is working on UNIX workstation, and necessary information to proceed the inference is transferred from the hospital information system using this modified HL7 message communication protocol. Less