Technical Research And System Implementation Of Data Retrieval For Similar Medical Records

With the popularization of information technology in the medical field,the rational and efficient application of these exponentially-growing medical data can greatly promote medical research,enrich medical applications,and serve people's daily lives.Medical record,with rich medical information and research value,is an important part of medical data.In this paper,through the analysis of large number of medical records,a similar medical record retrieval system is designed to realize the retrieval function of the overall medical record.The system is designed in three aspects.(1)In terms of medical record similarity measure,this paper designs a medical record similarity measurement algorithm based on weak supervised learning.Firstly,the method of multi-index probability allocation is used to construct the medical record group,which avoids the local optimal problem.Second,labels are weighted based on the theoretical model to make full use of theoretical knowledge.Finally,from a perspective of machine learning,the similarity of medical record is measured by the analysis of the loss function and the learning model.The algorithm combines a theoretical-based retrieval model with a data-based learning model to improve the accuracy of the case similarity measure.(2)In terms of cluster analysis of medical records,this paper proposes a medical records clustering algorithm based on fuzzy relations.First,based on the Spearman correlation coefficient,the medical record is processed to avoid the correlation error.Second,the posterior probability theory is introduced to weight stability,which reduces the influence of multi-source data of medical records.Finally,according to the principle of fuzzy transit closure,the medical records are clustered from the relationship transformation.The algorithm improves the accuracy of medical record clustering;simultaneously,it solves the dynamic and hierarchical problems.(3)In terms of retrieval feedback optimization,this paper proposes a method of medical record retrieval optimization based on multipath feedback.First,the system feedback is analyzed based on the environment used in the medical record retrieval system,with the summary and quantitative processing to improve applicability.Second,based on the implicit indirect feedback and explicit direct feedback of the system,rule-based label optimization feedback methods are proposed,respectively.Finally,a combined analysis of the two feedback methods is performed.From the actual environment of the system,this method improves the applicability and retrieval accuracy of the medical record retrieval system,with the self-optimization ability.In this paper,the similar medical record retrieval system optimizes the design of the three core modules based on the similarity measurement of medical records,the clustering of medical records,and the feedback retrieval of medical records.From theperspective of system,and the medical record retrieval system with the medical records as the query object is obtained by integrating modules.The system realizes the function of full-text search of medical records and the overall retrieval of similar medical records,and displays the medical record portrait and its statistical distribution in multiple dimensions.Different from the existing medical record retrieval system,the system designed in this paper realizes the similar medical record retrieval of the overall medical record input.The use of medical record data significantly improves the retrieval accuracy.The application of clustering is effective to accelerate the retrieval speed.The design of feedback information highlights the system's self-optimization ability.As a clinical auxiliary diagnosis and medical research tool,this system is inspiring for practical application and scientific research.