基于确定性因子理论的肺癌诊断Web专家系统的研究与实现

发布时间：2018-04-17 12:13

本文选题：肺癌诊断 + Web专家系统　；参考：《吉林大学》2017年硕士论文

【摘要】：现阶段,计算机人工智能研究领域在不断扩大,知识工程也在不断延伸,专家系统派生出来很多与之相关领域,如知识获取、知识表示、推理方法、不确定性和模糊性处理、专家系统开发工具和开发平台的研究等。虽然现在已开发的专家系统有很多,但是对于Web专家系统,目前Internet上可用的很少。肺癌的发病率和死亡率在不断增长,临床诊断的肺癌大部分已为中晚期。对于肺癌防治,需要早发现早治疗。普通人每4-6个月去医院检查一次才能达到早期诊断的目的,然而,这样防治措施实现起来难度很大。因此,可以借助专家系统模拟医生进行肺癌的早期初步诊断,将专家系统设计为B/S的模式,使得用户可以直接使用浏览器,通过Web上的肺癌诊断专家系统,进行早期肺癌诊断,明确自己的患病概率,为下一步的诊疗提供依据。在理论研究上,针对肺癌诊断的特点,本文提出了基于经典专家系统MYCIN中的不确定性处理方法确定性因子理论的改进模型。本系统模型中包括症状定性概率与定量概率的转化、定量概率与确定性因子理论中可信度的转化、对规则可信度的分段处理、加权“与”字段处理、规则拆分“或”字段处理、以及最终将最高假设可信度与肺癌疑似病患者吸烟信息综合,计算得出患病概率。本系统核心部分推理引擎基于JESS专家系统外壳,使用Rete模式匹配算法,与马尔可夫算法相比,本系统推理效率高。在技术研究上,由于JESS专家系统外壳是以Java语言为核心的,因此具有很强的扩展性。系统总体使用MVC设计模式,应用Java Web开发技术,前端使用JSP网页交互,服务器端使用Servlet调用底层的Java Bean和DAO完成数据库与知识库的访问,启动推理引擎获得推理诊断结果,并将诊断结果以诊断报告的形式返回给客户端。对于数据库连接,通过Tomcat数据源定义数据库连接池,应用JNDI查找的方式,提高了系统的整体运行效率。
[Abstract]:At present, the research field of computer artificial intelligence is expanding, and the knowledge engineering is also extending. Expert systems derive from many related fields, such as knowledge acquisition, knowledge representation, reasoning methods, uncertainty and fuzziness processing.Expert system development tools and development platform research.Although there are a lot of expert systems that have been developed, but for Web expert systems, there are few available on Internet.The incidence and mortality of lung cancer are increasing, and most of the clinically diagnosed lung cancer is in advanced stage.For lung cancer prevention and treatment, need early detection and early treatment.The average person goes to the hospital every 4 to 6 months to get the early diagnosis, however, the prevention and cure is very difficult.Therefore, the expert system can be used to simulate the doctor's early initial diagnosis of lung cancer, and the expert system can be designed as a B / S model so that users can use the browser directly and use the lung cancer diagnosis expert system on Web to diagnose early lung cancer.Identify their own probability of illness, for the next diagnosis and treatment to provide the basis.According to the characteristics of lung cancer diagnosis, an improved model of deterministic factor theory based on the uncertainty treatment method in classical expert system (MYCIN) is proposed in this paper.The system model includes the transformation of qualitative probability and quantitative probability of symptom, the transformation of reliability in the theory of quantitative probability and deterministic factor, the piecewise processing of reliability of rules, the processing of weighted "and" field, rule splitting or field processing.Finally, the highest hypothetical reliability and smoking information of suspected lung cancer patients were synthesized, and the probability of the disease was calculated.The inference engine of the core part of the system is based on the shell of JESS expert system, and uses Rete pattern matching algorithm. Compared with Markov algorithm, the reasoning efficiency of this system is high.In the technical research, the JESS expert system shell is based on Java language, so it has strong expansibility.The whole system uses MVC design pattern, Java Web development technology, front end uses JSP web page interaction, the server end uses Servlet to call the bottom Java Bean and DAO to complete the access of database and knowledge base, and starts the inference engine to obtain the inference diagnosis result.The diagnosis result is returned to the client in the form of diagnosis report.For database connection, the database connection pool is defined by Tomcat data source, and the overall running efficiency of the system is improved by using JNDI lookup method.
【学位授予单位】：吉林大学
【学位级别】：硕士
【学位授予年份】：2017
【分类号】：TP393.09

【相似文献】