Research And Application Of Data Mining Technology In Thyroid Disease Classification

With the rapid development of informationize medicine,many problems have been exposed in traditional medical methods,such as the waste of massive data and the untimely diagnosis and treatment.These problems are being gradually solved.Modern medical technology is constantly providing people with higher quality and higher efficiency medical services.With more and more clinical data being stored in medical database,applying data mining technology to patients’ disease data,mining valuable information,will make a very important contribution to medical decision-making and medical research.In recent years,researchers mainly apply data mining technology to chronic diseases such as diabetes,hypertension and cardiovascular disease,rarely in the direction of thyroid disease.Thyroid disease is a common high incidence disease in the field of Endocrinology,but also the second largest endocrine disease except diabetes.Therefore,we propose frameworks(AR-ANN and ARB)which combine association rules mining algorithms with machine learning classification algorithms in data mining technology,which can not only mine valuable information from massive information,provide scientific basis for prevention and control in the field of thyroid disease,but also effectively classify and diagnose thyroid disease.This paper designs and implements a data mining system for thyroid disease,which mines the knowledge hidden behind the actual clinical data of thyroid disease,and provides doctors with a strong basis for diagnosis.Two real thyroid datasets in UCI machine learning repository are used to analyze the frameworks,and the validity and correctness of the proposed frameworks are verified.First of all,data warehouse technologies such as data cleaning,transformation and integration are used to preprocess the two original datasets.Then the processed datasets are analyzed with multi-dimensional visualization.Next,taking the disease category and health category of thyroid disease as the post constraint conditions in association rule mining,two kinds of association rule mining algorithms(Apriori and Predictive Apriori)are used to mine the correlation rules between different attributes and study the correlation between the rules,so as to provide powerful help for clinical research of thyroid disease.At the same time,Apriori algorithm will also be used for feature selection and dimension reduction of datasets.Finally,the machine learning classification algorithms in data mining technology are divided into two categories,artificial neural network algorithm(BP neural network)and ensemble classification algorithm(Bagging classification algorithm),to classify and diagnose thyroid disease,and then we comprehensively analyze and compare the performance of each algorithm.Through comparative analysis,the optimal classification model is obtained,which is most suitable for the aided diagnosis of thyroid disease.At the same time,it also proves the feasibility and practical value of the frameworks in the aided diagnosis of thyroid disease.