Research On Medical Diagnosis Based On Sequential Three-way Decision

At present,the uneven distribution of medical resources and the uneven qualification of doctors make the misdiagnosis rate remain high.In order to improve the efficiency of doctors’ diagnosis,promote the medical service environment and actively respond to the "Internet + Medical" policy,it is necessary to build an intelligent support system to assist doctors in decision-making.In general,patient consultation is a process of gradually acquiring information.When the disease information is insufficient,it may cause huge losses.As a multi-stage decision-making method to deal with uncertain problems and in line with human cognition,the Sequential Three-way decision model can be applied to medical diagnosis.However,the traditional Sequential Three-way decision is limited to the equivalence relationship,and the decision thresholds are given artificially,which is easy to lead to decision conflict.In order to solve the above problems,the main research contents of this paper are as follows:(1)According to the characteristics of gradual acquisition of medical data,this paper constructs a Sequential Three-way decision diagnosis model based on K-nearest neighbor.Firstly,different similarity measurement methods are proposed for different numerical types of data.Secondly,referring to the idea of K-nearest neighbor classification,two conditional probability calculation methods θ-conditional probability and K-conditional probability are defined.Finally,the 0-1 loss function and three cost functions generated by delayed decision making are used as the basis for modifying the threshold value in the next stage,and the change rules of threshold value between different levels are explained.Experimental results show that the proposed model achieves good classification effect.(2)Aiming at the decision conflict caused by the change of granularity information from coarse to fine,this paper redefines the Sequential Threeway decision model under three different risk strategies,and constructs the Sequential Three-way decision diagnosis model with decision correction.Firstly,two kinds of positive and negative domain classification precision are defined for different granularity layers from local and global perspectives respectively.Secondly,according to the local classification precision difference of adjacent granularity layers and the relationship between local classification accuracy and global classification accuracy,the four categories of decision correction are discussed,and the corresponding decision correction rules are formulated.The model proposed in this paper can effectively improve the accuracy of diagnosis when applied to medical field.