Faculty Mentor
Dr. Ahmed Al-Taweel
Location
Russell Union Ballroom
Type of Research
Completed
Session Format
Poster Presentation
College
College of Science & Mathematics
Department
Mathematical Sciences
Abstract
In this study, we modified an epidemiological model of social media mob using deep learning methods to investigate information transmission using COVID-19 Twitter data from April to June 2021. Data collection was done using keywords, such as “covid 19”, “vaccine”, “lockdown”, and “quarantine” to simulate activity trends during the pandemic. In this work, we combine deep learning methods, including LSTMs and GRUs, with the susceptible-counter-infective-recovered (SCIR) model to forecast time-dependent parameters. Primary qualitative analyses, including the social media mob free equilibrium (MFE) point, the endemic equilibrium point, and the basic reproduction number ℜ0, were performed. Our analysis shows that the social media MFE point is locally asymptotically stable if ℜ0 < 1. The bifurcation existence and the stability of the steady states are established. Numerical experiments and sensitivity analysis of significant parameters are carried out. Finally, the results show that the proposed model outperforms pure deep learning and provides more reliable short and medium-term predictions.
Program Description
.
DOI
10.20429/GS4.2026.022
Start Date
4-23-2026 2:00 PM
End Date
4-23-2026 4:00 PM
Recommended Citation
Offei, Bismark Ampaw, "Coupling Deep Learning With Time-dependent Scir Model for Predicting Social Media Mob Behaviors" (2026). GS4 Student Scholars Symposium. 175.
https://digitalcommons.georgiasouthern.edu/research_symposium/2026/2026/175
Coupling Deep Learning With Time-dependent Scir Model for Predicting Social Media Mob Behaviors
Russell Union Ballroom
In this study, we modified an epidemiological model of social media mob using deep learning methods to investigate information transmission using COVID-19 Twitter data from April to June 2021. Data collection was done using keywords, such as “covid 19”, “vaccine”, “lockdown”, and “quarantine” to simulate activity trends during the pandemic. In this work, we combine deep learning methods, including LSTMs and GRUs, with the susceptible-counter-infective-recovered (SCIR) model to forecast time-dependent parameters. Primary qualitative analyses, including the social media mob free equilibrium (MFE) point, the endemic equilibrium point, and the basic reproduction number ℜ0, were performed. Our analysis shows that the social media MFE point is locally asymptotically stable if ℜ0 < 1. The bifurcation existence and the stability of the steady states are established. Numerical experiments and sensitivity analysis of significant parameters are carried out. Finally, the results show that the proposed model outperforms pure deep learning and provides more reliable short and medium-term predictions.
