Forecasting virus outbreaks with social media data via neural ordinary differential equations

Abstract

In the midst of the covid-19 pandemic, social media data collected in real time has the potential of being an early indicator of a new epidemic wave. This possibility is explored here by using a neural ordinary differential equation (neural ODE) that is trained to predict virus outbreaks for a geographic region. It learns from multivariate time series of signals obtained from a novel set of massive online surveys about COVID-19 symptoms. Once trained, the neural ODE is able to capture the dynamics of the interlinked local signals and accurately predict the number of new infections up to two months in advance. Moreover, it can estimate the future effects of changes in the number of infected at a given time, which can be associated with the flow of people entering or leaving a given region or, for instance, with a local vaccination campaign. This work gives compelling preliminary evidence for the predictive power of widely distributed social media surveys for public health application

Hybrid deep learning of social media big data for predicting the evolution of COVID-19 transmission

“…Besides comparing ODANN with other time-series prediction models as explained qualitatively and quantitatively in the preceding sub-section, we further validate the predictive accuracy of ODANN with recent similar notable studies [42] , [43] , [44] , [45] , [46] , which too focused on forecasting the transmission rate of COVID-19, in terms of the number of confirmed COVID-19 cases, since the virus’ inception. In the following, we outlined the key methodologies and reported results by the previous studies [42] , [43] , [44] , [45] , [46] , for comparison with our proposed ODANN model and its prediction results.…”

“…A more similar study in using social media to forecast the virus outbreak with neural ordinary differential equations (ODEs) was performed by Núñez et al. [44] . The authors’ data comprised of a massive amount of online surveys, regarding COVID-19 symptoms, via Facebook to train and validate the authors’ personalized neural ODE, followed by using the trained neural ODE to forecast the virus’ outbreak rate in different US states for up to sixty days.…”

Using scientific machine learning for experimental bifurcation analysis of dynamic systems

“…We carried out numerical simulations using the 2+1 DoF flutter model given in Eq. (10), to obtain its bifurcation diagram. We studied both the deterministic scenario and the case when the system was polluted with additive noise.…”

“…Several studies have used UDE or NDE models for physical structures [8], pandemic- [9,10] and climate modelling [11]. It is worth highlighting the work of Rackauckas et al [12] to create an environment in the Julia programming language which allows for construction and training of such models.…”

Socio-economic pandemic modelling: Case of Spain

“…), upo which we could do data-driven modeling, as there is trustworthy quantification of the number of contagions and deaths caused by Covid-19. There are various models that have recently been developed to deal with the dynamical spread of the disease over a geographical region, (see for instance [3][4][5][6][7][8] ), and also studies about the behaviour of several economic indexes during hazardous times 9 . However, only a few studies take into account both, the economic performance and the spread of the pandemic simultaneously, which is the problem we focus on here.…”