Research And Application Of Rough Set Theory In Dealing With Massive Electronic Medical Records

With the development of the wisdom of health care,a large number of medical data resources were integrated together.As a valuable asset,a lot of medical data mining had become a research focus in the current academic field.Because of the increase of data quantity and redundant attributes,it was difficult to excavate knowledge mining.How to reduce the mass medical data effectively and improve the efficiency of knowledge mining was the research direction of this paper.Rough set theory was very powerful in the research of incomplete data,inaccurate knowledge representation,generalization,learning and so on,and the attribute reduction was one of the main applications of the theory.In this paper,the problem of rough set attribute reduction algorithm was summarized,an optimization strategy and a parallelization scheme combining rough set attribute reduction with tabu search algorithm were proposed,the performance of the algorithm was verified by simulation experiments and disease classification experiments,it not only provided a good idea for the improvement of the reduction algorithm,but also provided the possibility for the efficient processing of large data sets.The specific research contents were as follows:(1)Common rough set attribute reduction algorithms were analyzed by consulting relevant domestic and foreign literature,not only the problems between the algorithms were summarized,but also the main content of this paper was determined.(2)According to the characteristics of rough set theory and tabu search algorithm,the attribute reduction algorithm based on tabu search was proposed.Firstly the composition of the algorithm was described,which included representation of solution,solution precision metric,tabu list,generation of neighboring candidate solutions,universality and concentration patterns,and the whole implementation process of the algorithm was introduced.In order to improve the expansibility of attribute reduction algorithm of tabu search,the parallelization scheme of attribute reduction algorithm based on tabu search was proposed.(3)In order to test the basic performance of attribute reduction algorithm based on tabu search,the UCI data set was used as the experimental data,the proposed algorithm and several common attribute reduction algorithms were used to carry out simulation experiments,accordingto the experimental results,the feasibility,stability and reduction effect were compared and analyzed.(4)In order to test the effectiveness of attribute reduction algorithm based on tabu search,Hadoop experimental environment was built,massive electronic medical records were used as experimental data,in the data preprocessing stage,the traditional four attribute reduction algorithms and the attribute reduction algorithm based on tabu search proposed in this paper were used to attribute reduction,five kinds of disease classifiers were constructed by using naive Bias classification algorithm in classification stage.Through the disease classification experiment,the effectiveness of the attribute reduction algorithm based on tabu search was proved.