Epidemiological models in health misinformation propagation of Mpox conversations

The rapid dissemination of health misinformation through digital platforms poses significant challenges to public health communication and response strategies. The spread of misinformation can be likened to the spread of diseases within communities during an epidemic. This study addresses the question around how the classical SIR (Susceptible-Infectious-Recovered) and adapted SEIR (Susceptible-Exposed-Infectious-Recovered) epidemiological models compare in their abilities to capture the temporal dynamics of health misinformation spread on social media platforms during disease outbreaks. Using the 2022 Mpox outbreak as a case study, we analyse a publicly available dataset of 5,787 tweets collected between 7–13 July 2022, with the vast majority posted on a single day (9 July 2022). Given the narrow temporal window and concentrated activity in this dataset, we frame our study as a proof-of-concept illustrating how compartmental models can be adapted to short-horizon social media bursts, rather than a definitive epidemiological characterisation of Mpox misinformation. Within this scope, our comparative analysis shows that the SIR model effectively captures rapid initial dissemination patterns characteristic of viral misinformation spread, whilst the SEIR model more accurately represents the latency period between exposure and active sharing behaviour observed in social media contexts. We report explicit parameter values ( β , γ , σ ), discuss their sensitivity and identifiability limits under the short observation window, and highlight the restrictions imposed by the homogeneous-mixing assumption. These findings contribute methodological insights for compartmental model selection in misinformation research and motivate future work with larger, longitudinally collected datasets.