Research Of Multi-Agent Based Clinical Knowledge Representation With Its Dynamic Parse And Execution

With the development of evidence-based medicine, digital medical and a series of technologies, medical information has gradually become an important development direction in the field of contemporary medical health. However, as China is a large country with different economic conditions between the east and west, the process of medical information is relatively slow. At present, non-standardization is an important factor of hindering the development of medical information. During the operation of HIS products, problems of non-standard communication interface are serious, which results in low-level HIS, repetitively developed and difficult to transplant, to promote and to share data. Clinical guidelines(CGLs) are a set of clinical knowledge based on the evidence-based clinical guidelines which are accepted by a large number of doctors and patients. Using the CGLs can help to increase the accuracy and security of clinical decisions. Thus, they are of great guiding importance to the clinical decisions that the doctors make during the process. However, because the clinical guidelines are usually described by simple text and format, it is difficult for the computer to operate and progress. Due to a lack of unified consistent expression for a large number of clinical knowledge in a distributed environment, clinical knowledge cannot effectively provide the doctors valuable decision-making service. Thus, we can not wait to solve the problem by making the clinical guidelines standard.This paper aims to study how to convert clinical knowledge which are accepted by medical experts and patients into knowledge structure of computer executable. Based on clinical guidelines, the doctors inquire and collect necessary and sufficient information from patients in the process of clinical decision support. Through the clinical analysis of knowledge and the execution engine, standardization, standardized clinical knowledge are applied to the clinical decision support system, then assisting doctors making medical decisions rapidly, efficiently and accurately. According to the research content, we make a discuss about this paper from the following aspects:(1) Definite a set of tag to express the clinical knowledge; According to the label set document, develop a clinical knowledge acquisition tools(developed by other team members); With clinical knowledge acquisition tool, make the clinical knowledge XML, then check by the clinical knowledge acquisition tool, and if the result is correct, deposit it into the clinical knowledge base; If there are any errors, redefine and convert it according to the prompt.(2) Design and implement the dynamic analysis and execution engine of clinical knowledge to perform structured clinical knowledge. By matching the symptoms of patients and clinical knowledge, calculate the value of the decision options and give the optimal decision.(3) Design and implement dynamic interface generation engine based on rules, with the engine’s analysis and combination of clinical knowledge documents and electronic medical records, dynamically generate the sufficient and necessary inquiries information of the patients during the process of the clinical decision making, and display the information in the form of a page, thus completing the dynamic and unrepeated work of information collection of patients.(4) By introducing the technology of Multi-Agent, the dynamic inquiry and clinical knowledge based on rules are embedded into the dynamic execution in a distributed environment, so as to realize the asynchronous of diagnosis, treatment and cooperation between different departments.This paper is based on the triple assessment for the case of breast cancer, verify the effectiveness and rationality of representation, dynamic parse and execution of Multi-Agent based clinical knowledge. This paper proves the versatility and portability between clinical knowledge and dynamic analytic according to the case about the treatment of head injury.