基于武陵山片区苗族医学胃病诊断理论的临床决策系统研究

临床决策系统是人工智能在医疗领域的重要应用之一。近年来，国内基于民族医学（如藏医、蒙医、维医）的临床决策系统研究正在顺利进行，但基于苗族医学的临床决策系统研究尚处于起步阶段。针对目前的情况，本课题以武陵山片区苗族医学胃病诊断相关资料为基础，拟采用一种改进的模糊Petri网模型来实现基于基于武陵山片区苗族医学胃病诊断理论的临床决策系统。具体而言，本课题包括如下三个方面的研究内容：1、将改进和声搜索算法与深度学习理论相结合提出一种知识抽取模型，并将此模型运用到基于武陵山片区苗族医学胃病诊断知识库的获取过程中；2、根据实际诊断过程中的“初诊-确诊”的循环诊断模式，提出一种含多阈值变迁的多分支带环路的改进模糊Petri网模型，并实现基于该改进模糊Petri网的胃病诊断知识库等价建模；3、针对“状态爆炸”问题，提出一种基于可达树的大规模改进模糊Petri网模型分解算法，并实现基于最佳子改进FPN模型

The clinical decision support system (CDSS) is one of the most important application domains of artificial intelligence in the medicine area. In China, CDSS has employed to execute the disease diagnosis rooted in some ethnomedicines, such as Tibetan medicine, Mongolian medicine, and Uyghur medicine. However, compared to these successful applications of CDSS in mentioned ethnomedicines, the study of CDSS based on Miao medicine is still in the preliminary stage. Focuses on this situation, this project tries to propose CDSS to automatically perform the gastropathy diagnosis function based on the collected gastropathy diagnostic knowledge-base of the Miao medicine theory of the Wuling mountain area. To fulfill the predetermined goal of this project, the following three main issues will be discussed, which are: (1) to obtain the gastropathy diagnostic knowledge-base of the Miao medicine theory of the Wuling mountain area by using a hybrid modified HS-DL knowledge extraction framework; (2) to propose a modified formalism of fuzzy Petri net (FPN), namely multi-branches and loop structure FPN with the multi-threshold transitions after analyzing the process of disease diagnosis and to present an equivalent transformation approach between the gained gastropathy diagnostic knowledge-base and the extension FPN model; (3) to discuss a decomposition algorithm of the large-scale modified FPN for overcoming the ‘state explosion’ issue using reachability tree and to realize the automatic gastropathy diagnosis function based on the optimal sub- modified FPN model.