Analysis And Mining Of Clinical Data Based On Pathological Nipple Discharge

Because of the differences of region,environment,life style,cultural background and medical and health conditions,the incidence of breast disease is different in different regions.For breast diseases,nipple discharge is one of the common symptoms next to breast pain and breast mass.At present,there is no unified and standardized diagnosis and treatment guidelines for pathological nipple discharge at home and abroad,so it is very important to improve the diagnostic accuracy of pathological nipple discharge.This article attempts to get the epidemiological characteristics of pathological nipple discharge through the observation of pathological nipple discharge under the different nature of mammoscopy in Liuzhou area,so as to provide the basis for the diagnosis of nipple discharge and reduce the incidence of further malignant development of pathological nipple discharge to breast cancer.Based on the clinical data of 1219 women with breast disease from 2013 to 2018 in Liuzhou City,this paper has a basic understanding of the data through descriptive analysis.In the modeling stage,the data is preprocessed first.In the establishment of association rule model,the maximum change of association degree is determined according to the support degree,confidence degree and promotion degree by correlating four diagnostic variables with the rest variables respectively The amount of nipple discharge,and then get the diagnosis of non-tumor nipple discharge is more likely to be related to the situation.Finally,the variables of "diagnosis result" were treated as three 0-1 dependent variables: "whether it is a tumor occupying place","whether it is an inflammation of the mammary duct" and "whether it is an expansion of the mammary duct".The logistic regression model was established with the remaining variables as independent variables,so as to obtain the relevant clinical characteristics of various diseases.Through association rule mining technology,when the clinical manifestation is no mass,the overflow character is serous and the overflow color is yellow,it is often diagnosed as non tumor nipple overflow;the establishment of logistic regression analysis model analysis shows that when the clinical symptoms are porous,the overflow color is colorless or brown or red,the overflow character is blood and the presence of occupation In order to determine whether it is a tumor nipple discharge,it is more necessary to detect the location of the lesions by mammoscopy.The epidemiological characteristics of pathological nipple discharge in Liuzhou were summarized.This is helpful for doctors to accurately diagnose patients with nipple discharge.