Abstract:In order to improve the diagnosis accuracy, machine learning method was proposed to construct the breast cancer diagnosis model. The parameters of cell feature are the inputs of model and the class of diagnosed cell is the output. Three machine learning methods are chosen as training algorithm, including BP neural network, learning vector quantity network and support vector machine. Simulation results show that three methods have high identification ability (BP:97.28%, LVQ: 98.06%, SVM: 98.45%) and can be applied to other medicine research as effective method.