Fusing Tweets, Confirmed Cases, and Death Data to Accurately Now-Cast for COVID-19

• Activity Number: 41 - Storytelling on COVID-19 Impact Using Experts' Prior Knowledge and Data from Social Media, Official Clinical Data, Digital Phenotype from Smartphones' Raw Sensor Data, and Emergency Departments
• Type: Invited
• Date/Time: Sunday, August 8, 2021 : 3:30 PM to 5:20 PM
• Sponsor: International Society for Bayesian Analysis (ISBA)
• Author(s): Conor Michael Rosato* and Matthew Carter and Robert Moore and John Heap and Simon Maskell and Jose Storopoli
• Companies: University of Liverpool and University of Liverpool and University of Liverpool and University of Liverpool and University of Liverpool and UNINOVE - Sao Paulo - Brazil
• Keywords: Markov Chain Monte Carlo; SEIRD; Bayesian Statistics; Social Media; Now-casts; COVID-19

Abstract

The emergence of the novel coronavirus (COVID-19) has generated a need to quickly and accurately assemble up-to-date information related to the current spread. While it is possible to use deaths to provide a reliable feed of information, the latency of data derived from deaths is significant. Confirmed cases, derived from positive test results, will potentially provide a lower latency data feed. We propose to use machine learning (in each of multiple languages) to process Tweets’ text to identify which Tweets relate to symptomatic individuals and to use geolocation information, where available, to infer counts of symptomatic individuals Tweeting in each country or region each day. We then use an extended SEIRD model to fuse social media data, confirmed cases and deaths to estimate parameters of the model and now-cast the number of people in each compartment within the model. It is built using a system of ordinary differential equations (ODEs) using the probabilistic programming language Stan. Results are anticipated to show that the use of Twitter counts narrows the confidence bounds around these estimates when compared with using mortality data and/or confirmed cases alone.