Application Of Outlier Detection In The Abnormal Analysis Of Medical Prescription

In recent years,the medical accident is endless,the contradiction between doctors and patients is becoming increasingly acute,the medical dispute is more and more.There are many reasons for this situation,for example,the doctor's personal medical level is limited,doctors abuse drugs,unknown risks in drugs and so on.In order to relieve these problems,at the same time to reduce the economic burden of patients and cure the patients with disease quickly,in this paper,outlier detection algorithms can be used to scientifically analyze medical prescriptions of medical institutions.Based on the above background,this paper proposes a novel outlier detection algorithm based on the dissimilarity of the nearest set,and improves a outlier detection algorithm based on the isolation of the nearest set.First,the medical data is preprocessed,including the deletion of no actual meaning attributes and incomplete medical data.And then obtains doctor information data,patient information data,drug information data,medical prescription data,and the like through medical data.Reduce the dimension of the medical prescription data through the division of departments.Reduce dimension again by deleting the column vectors which are zero in the matrix of the medical prescription data.Finally,the outlier detection algorithm based on the dissimilarity of the nearest set and the outlier detection algorithm based on the isolation of the nearest set are used to mine and analyze the department medical prescription data with high dimension and large data.The outlier detection algorithm based on the dissimilarity of the nearest set calculate the degree of dissimilarity between data points by the attribute dissimilarity.The algorithm can deal with the data whose attribute is associated more efficiently and the time complexity of the algorithm is lower.the outlier detection algorithm based on the isolation of the nearest set more quickly detect the isolated data points from medical prescription data.However,this algorithm does not consider the correlation between drugs and cannot analyze the medical prescription data effectually.Apply actual medical data of a hospital,tests the proposed outlier detection algorithm based on the dissimilarity of the nearest set can more effectively and quickly mine the abnormal medical prescriptions from the medical prescription data.