Construction Of Hierarchical Decision-Making Knowledge Map For Hemophagocytic Tissue Syndrome

Hemophagocytic lymphohistiocytosis(HLH)is a serious clinical disease which seriously threatens the life and health of patients.As HLH hematopathy is a clinical medical syndrome with multiple causes,the clinical manifestations and laboratory examination results of HLH patients are different.If HLH-94 and hlh-04 are used in the same way,patients often die of infection in the course of treatment.Therefore,according to the different clinical manifestations and laboratory examination results,according to the severity of HLH patients with hematological disease,different intensity of treatment were used.Through the construction of HLH blood disease knowledge map and the development of clinical hierarchical treatment decision-making system to assist doctors in medical decision-making.According to a specific clinical manifestation,the targeted treatment plan is given.In this study,the hlh-2018 protocol(cchg-hlh-2018)and the treatment protocol in the electronic medical records of HLH patients were used as the treatment criteria.The pre training language model of Bert is mainly composed of attention mechanism and transformer model.By identifying the input sequence and the context information of the text,it can effectively solve the polysemy in the text.Then,blstm model is used to encode the word vector sequence of Bert.Finally,the CRF model is used to calculate the optimal tag sequence based on the output of blstm layer.According to the optimization algorithm,the entities related to HLH blood disease treatment in the original data set are identified to form the entity set of HLH blood disease knowledge map.The experimental results show that the accuracy of entity recognition is 92.35%.Then,under the guidance of experts and doctors,the HLH blood disease guidelines are analyzed to get the concept relationship,and the obtained concept relationship is filled into the concept pattern,so the HLH blood disease knowledge map schema(concept layer)structure pattern is designed and completed.Then,the entity set of HLH blood disease is mapped to the schema layer,and the HLH blood disease knowledge map is constructed.Based on the time axis,the map of examination items before treatment,the map of treatment plan during treatment and the map of curative effect evaluation after treatment were constructed,and the three were linked to form a HLH hematology knowledge map.Then the constructed atlas is applied to the hierarchical decision-making system of adjuvant therapy.The system can get the corresponding treatment plan stored in the atlas according to the pre-treatment examination results,and assist doctors to make medical decisions.