Research And Application Of Medical Diagnostic Model Based On Electronic Medical Record

With the rapid development of Internet technology and the continuous advancement of medical informatization,electronic medical records(EHR)has gradually replaced complicated handwritten medical records and become an indispensable part of the hospital.The emergence of EHR is of great significance for the development of medicine.It provides not only convenience and high efficiency for the health management of hospitals,but also data support for further research in medicine.However,the potential value of EHR data has not been fully excavated because of its large scale and complex structure.Therefore,aiming at the problem of low utilization of big data in EHR,this paper focuses on the medical diagnosis model based on the electronic medical record data,and finally realizes the intelligent diagnosis of the disease,in order to provide medical staff with effective reference,improving the efficiency and accuracy of medical diagnosis.The main contents of this paper are as follows:(1)By consulting a large amount of relevant literature and knowing the domestic and foreign research status,this paper analyzed and studied the medical diagnosis process,summarized the existing problems,and determined the research direction and main contents of this paper.(2)In the medical diagnosis process,there are more redundant and weakly related features in disease data,which reduces not only the efficiency of diagnosis model construction,but also the accuracy of disease diagnosis.Therefore,this paper chose the feature selection algorithm based on Particle Swarm Optimization(PSO)to reduce the feature of the disease,and proposed an improved RS-BPSO feature selection algorithm based on the existing problems in the original algorithm.Finally,through the experimental simulation of UCI dataset,the rough set algorithm and the binary particle swarm optimization algorithm were compared to verify the advantages of RS-BPSO algorithm in feature attribute reduction,classification accuracy and convergence time.(3)In view of the disadvantages of the existing medical diagnostic models in the diagnostic accuracy,a BN-DS two-level diagnosis model combining Bayesian network with DS evidence theory was proposed in this paper.As the same,the diagnostic process and steps were described by the flow chart.First,a number of Bayesian sub-network model were established for the initialdiagnosis of disease.Then the preliminary diagnosis results were fused to get the final diagnosis results through the DS evidence theory.(4)In order to verify the validity of the BN-DS diagnostic model,the patient's electronic medical records were used as experimental data for diagnostic experiments in this paper.Firstly,the text of electronic medical record was preprocessed,including text segmentation,stop words and feature extraction.Then the feature attributes were selected by the RS-BPSO algorithm proposed in this paper.Finally,the BN-DS secondary diagnosis model was used to diagnose the disease.Through the comparison and analysis of the experimental results,the model was validated to effectively improve the accuracy of disease diagnosis.