Research On Breast Cancer Diagnosis Method Based On Support Vector Machine

Breast cancer is a disease that seriously harms women’s health.With the change of lifestyle,the incidence rate has increased significantly,and the age of onset will also tend to be younger.Early detection,early diagnosis,and early treatment can reduce the possibility and fatality rate.In the actual medical detection system,due to the limitation of medical equipment or the influence of human physiological factors,misdiagnosis or missed diagnosis of breast cancer is caused,which further makes the patient lose the opportunity of early detection and early treatment.Therefore,improving the accuracy of breast cancer diagnosis is an important means to reduce the life risk of patients.With the development of artificial intelligence,using intelligent data mining methods to mine the valuable medical data is a new way to achieve intelligent diagnosis.In this paper,breast cancer,a common disease threatening women’s health,is selected as the research object of algorithm application.The method of data mining is used to model and analyze the data of breast cancer.Based on the clinical breast cancer data provided by UCI,a support vector machine aided diagnosis model is established.By optimizing the model,the accuracy of breast cancer diagnosis is improved.A new research idea and method for intelligent medical treatment is proposed,which has certain practical significance and application value.The main contents are as follows:(1)Analyze the main intelligent medical treatment and the application of machine learning algorithms in intelligent medical treatment,learn the support vector machine theory and linear and nonlinear principles,and implement high-dimensional linear transformation by introducing kernel functions,and then propose optimization problem of support vector machine kernel functions Parameter.(2)A support vector machine based on breast cancer diagnosis research method based on parameter optimization is proposed.In view of the two key factors that affect the classification of SVM: error penalty parameter C and kernel function form and its parameter problem,a parameter optimization scheme is proposed.The optimized grid search method and crossvalidation method are used to optimize the parameters,effectively avoiding the genetic algorithm,Particle swarm optimization,and other local optimization problems.The experimental results show that the method is beneficial to improving the performance of the classifier.Compared with the traditional method,training takes less time and the accuracy is significantly higher.(3)The support vector machine method based on Relief-F feature weighting is proposed and used for breast cancer data modeling and analysis.In view of the defect that the current support vector machine only considers the importance of samples and ignores the influence of feature importance on the classification results,a support vector machine method based on Relief-F feature weighting is proposed.This method first uses Relief-F feature weighting method to calculate each feature and then uses the obtained feature weight value to calculate the inner product and Euclidean distance in the kernel function,so that the calculation of kernel function is not dominated by some weakly correlated or uncorrelated features.The results of theoretical analysis and numerical experiments show that this method has better robustness and classification ability than traditional SVM.