Consejo Nacional de Investigaciones Científicas y Técnicas (CONICET), Buenos Aires, Argentina.
Departamento Materiales Nucleares, Centro Atómico Bariloche, Comisión Nacional de Energía Atómica (CNEA), Bariloche, Argentina.
Sci Rep. 2023 Jul 5;13(1):10870. doi: 10.1038/s41598-023-37118-9.
During the Covid-19 pandemic, real-time social media data could in principle be used as an early predictor of a new epidemic wave. This possibility is examined here by employing a neural ordinary differential equation (neural ODE) trained to forecast viral outbreaks in a specific geographic region. It learns from multivariate time series of signals derived from a novel set of large online polls regarding COVID-19 symptoms. Once trained, the neural ODE can capture the dynamics of interconnected local signals and effectively estimate the number of new infections up to two months in advance. In addition, it may predict the future consequences of changes in the number of infected at a certain period, which might be related with the flow of individuals entering or exiting a region. This study provides persuasive evidence for the predictive ability of widely disseminated social media surveys for public health applications.
在新冠疫情期间，实时社交媒体数据原则上可以用作新疫情浪潮的早期预测指标。本研究通过使用经过训练以预测特定地理区域病毒爆发的神经常微分方程（neural ODE）来检验这种可能性。它从有关 COVID-19 症状的一组新颖的大型在线民意测验中得出的多元时间序列信号中进行学习。经过训练后，神经 ODE 可以捕获本地相互关联信号的动态，并有效地提前两个月估算出新的感染人数。此外，它可以预测在特定时期内感染人数变化的未来后果，这可能与进出一个地区的个人流动有关。这项研究为广泛传播的社交媒体调查在公共卫生应用中的预测能力提供了令人信服的证据。
