BiLSTM Deep Learning Model for Heart Problems Detection

Jakub Silka, Michal Wieczorek, Martyna Kobielnik, Marcin Wozniak

Abstract: Deep learning architectures find applications where analysis of complex data inputs is demanding and regular neural networks may have problems. There are many types of deep learning models, however the most important to fit architecture and training model to the input data. In this article we propose a model of deep learning based on architecture in which we use BiLSTM neural network. Proposed model is trained by using Adam algorithm. For the research experiment we have examined also other latest algorithms to select the best configuration of proposed model. Results show that our proposed BiLSTM deep learning neural network archived over 99% of accuracy.