An Integrated Climate and Spatio-temporal Determinant for Influenza Forecasting based on Convolution Neural Network

Abstract

We have proposed convolution neural network (CNN) for influenza forecasting. Our experiment model provided time series datasets consisted of climate variables and the spatio-temporal forms. The climatic variables have included precipitation, snowfall, maximum temperature, and minimum temperature. The spatio-temporal procedure has had two features, a flu feature is the influenza patient count in different time of focus region node. The second feature is the influenza patient count from influenza carrier in adjacent region. We assumed that asymptomatic patients which is a carrier of influenza. They will be able to travel anywhere whenever needed on pedestrians, vehicles or planes in their positions. We have provided two effect flu factors in climatic and human into deep machine learning for accurate predictions of influenza outbreaks results. The integrated variables influenced effectively influenza node predictions. The research compared models on recurrent neural network (RNN), long short-term memory neural network (LSTM) and convolution neural network (CNN). The term of Mean Absolute Percentage Error (MAPE), Root Mean Square Error (RMSE), and Root Mean Square Percentage Error (RMSPE) captured for evaluate model. The performance denoted following resulted the convolution neural network (CNN) combined with Integrated climate and spatio-temporal determinant. CNN approved significant influenza forecasting more effectively than recurrent neural network (RNN) and long short-term memory neural network (LSTM).