Abstract
Disease transmission modeling involves complex spatio-temporal patterns, considering not only the disease transmission in the temporal dimension but also the impact of population migration between different locations. We propose an Intra-locational Migration Estimation-based Graph Neural Network (IME-GNN) that is different from traditional time series forecasting. In this model, nodes represent locations, and edges represent population migration between locations. In IME-GNN, temporal module and spatial module are used to model the transmission of diseases in temporal and spatial dimensions respectively. In the absence of data on population migration within locations, IME-GNN achieves better results than the baselines by estimating population migration within locations. The model is applied to the dataset of real confirmed cases in Spain. The experimental results show that under limited data conditions, through the estimation of population migration within locations, IME-GNN can better learn the complex spatio-temporal dynamics of disease transmission.
Published Version
Talk to us
Join us for a 30 min session where you can share your feedback and ask us any queries you have
Schedule a call