Prediction of Hospital Outpatient Volume Based on Combined Neural Network

Abstract

The hospital outpatient volume has temporal property, and the importance of features is different. In this paper, a new neural network model was proposed in order to the regression prediction of outpatient volume, which adds attention layers to the original network model. In our paper, we use the powerful feature extraction ability of convolution neural network(CNN) to extract important features of data, and the attention mechanism is added in CNN, it helps the model to solve the blindness problem of CNN feature extraction, so that our feature extraction model can gives more sources on the important features and weaken the influence of other features. We connect the output of CNN to the input of long short-term memory network(LSTM), and add an attention layer to the LSTM hidden layers in order to better learn the time characteristics, the output of the last time step of LSTM is connected to a deep neural network to predict the results. The comparison with common algorithms shows that the error of the algorithm proposed in this paper is smaller.