Research On Diagnosis Model Of Breast Cancer Based On Integration Learning

At present,breast cancer is one of the most frequent malignant tumors in the world,and its development has greatly endangered the life and health of the women.Common breast cancer detection methods include X radiation,CT,thermography,ultrasound imaging and other methods.However,the above methods not only require high cost of inspection,but also bring great harm and pain to patients.If a low cost,high efficiency and small injury diagnosis method is applied to early diagnosis,it can reduce patient's pain and reduce its economic expenses.Therefore,the establishment of breast cancer diagnosis model has high practical value.By using the established breast cancer detection model,we can check the suspected breast cancer patients,so as to assist doctors in clinical decision-making and improve the early diagnosis rate of breast cancer.Based on the three assessment of breast cancer in the laboratory,this paper proposes a machine learning method for the diagnosis of breast cancer.Machine learning data sets from the UCI machine learning database in 699 groups of breast cancer,breast cancer diagnosis index of the relevant personnel data set for the University of Wisconsin medical research center extraction,through stepwise regression analysis and principal component analysis method to analyze the PCA data set of 10 attributes related to breast cancer,in order to get attributes associated with breast cancer,and keep it as input variables of BP neural network,decision tree,integration learning machine learning.Early diagnosis of breast cancer not only brings great pain to the patients,but also has a higher diagnostic cost.Machine learning has higher accuracy when dealing with more complex problems,and has a good prediction effect for new samples,so that the machine trained model can assist clinicians to diagnose and improve the early diagnosis rate of breast cancer.This paper will use the data of breast cancer diagnosis in UCI data set as the experimental data of this paper,and divide 683 sets of data(16 groups are incomplete data),and divide them into training data set and test data set according to the ratio of 6:4.Then,we use machine learning C4.5 decision tree,BP neural network and ensemble learning model to build corresponding disease diagnosis models respectively.Finally,we use the test data to test all the established algorithm models.The experiments show that the prediction results of each model are strongly correlated with the original data,which indicates that the prediction effect of the established model is better.The algorithm based on BayesNet,Logistic,DecisionTable and other algorithm integration is not only better than any one base classifier,but also has better classification ability than the commonly used BP neural network and C4.5 decision tree algorithm.Therefore,the final analysis and comparison of the integration learning model I proposed as the best model for the study of breast cancer diagnosis.