Back Propagation Neural Network Based Stroke Prediction

Abstract:
Stroke prediction is crucial to help doctors or specialists’ diagnosis the survival rate of patients. To help people prevent stroke, which has a high fatality rate, this paper preprocessed a dataset got from Kaggle website, then did the feature selection process by constructing decision tree model, and a deep learning based back propagation neural network (BPNN) model was built. After that, the cleaned data were fed into the model for training and testing to obtain results. Meanwhile, some experiments are done for selecting the most appropriate hyper parameter of the model, including hidden layer number and the neuron nodes number. Finally, the neural network model reached 98.13% training accuracy and 97.97% testing accuracy, which shows that the back propagation neural network model this paper constructed is feasible and fit for stroke prediction in this study. More datasets will be considered in the future for further verification based on our proposed methodology.