Epidemiological Models for Information Diffusion

27 concepts26 papers1 proposal

When we describe a video as "going viral" or a rumor as "spreading like wildfire," we are using biological metaphors for social phenomena. However, research across social media analysis, network engineering, and cybersecurity suggests that these are not just metaphors, but precise mathematical descriptions. By treating information as a pathogen and users as hosts, researchers can apply the tools of epidemiology to predict, manage, and curtail the spread of digital content.

The Compartmental Framework

The core intuition behind this approach is that information spread is a contact-driven process. In a biological epidemic, an individual’s state changes based on their interaction with others. Epidemiological diffusion modeling maps these transitions to digital behavior using "compartments."

In the classic SIR framework, a population is divided into three states:

  1. Susceptible (S): Users who have not yet encountered the information but are capable of being "infected."
  2. Infected/Active (I): Users who have consumed the content and are currently spreading it to their peers.
  3. Recovered/Removed (R): Users who have seen the content but have stopped sharing it, either due to loss of interest or encountering corrective information.

Christian Bauckhage et al. demonstrate in How Viral Are Viral Videos? that the attention dynamics of YouTube videos closely mirror these biological outbreaks. By calculating an "infection rate," the authors show that viral lifecycles are predictable: a rapid spike in "infections" (views) is followed by a decay as the pool of susceptible users shrinks and active spreaders move to the "recovered" state. Similarly, Zhi Wang uses this framework in Data-driven Approaches for Social Video Distribution to model how social re-sharing turns passive consumption into a viral cascade.

Modeling the Lifecycle of a Rumor

While simple SIR models work for videos, rumors and misinformation often require more nuanced states. Yichao Zhang 0001 et al. introduce the concept of "stiflers"—users who know the rumor but choose not to spread it—and "modifiers" who change the content as it passes through them. In Rumor evolution in social networks, the authors use differential equations to show how these behaviors cause rumors to simplify or morph into misinformation over time.

The "recovery" state is particularly critical in the fight against "infodemics." Leon Derczynski and Kalina Bontcheva argue in Pheme: Veracity in Digital Social Networks that users transition to a recovery state when they encounter corrective data. By modeling this as a Markov process, they can simulate how quickly a rumor dies out when challenged by verified facts. Shuiqiao Yang et al. applied these principles to COVID-19 myths in Analysis and Insights for Myths Circulating on Twitter During the COVID-19 Pandemic, identifying "fear" as the primary driver that increases the "reproduction number" ($R_0$) of misinformation.

Why Social Epidemics Stop

If information spreads like a virus, why doesn't every interesting tweet reach the entire world? Greg Ver Steeg et al. address this in What stops social epidemics?. They identify two primary dampening forces:

  • Network Clustering: Information often gets trapped in tight-knit communities where everyone already knows the "story," preventing it from jumping to new susceptible clusters.
  • Friend Saturation: Unlike biological viruses, where repeated exposure increases the chance of infection, social media users often become less likely to share something if they see it too many times from different friends.

Thinh Minh et al. further refine this in Non-linear Time-series Analysis of Social Influence by adding a "vigilant" state, representing users who become immune or indifferent to a trend after the initial shock wears off.

Engineering Applications

Beyond social science, these models provide a rigorous framework for optimizing technical systems:

By abstracting the specific content—whether it is a YouTube video, a COVID-19 myth, or a blockchain transaction—into these mathematical states, researchers can move from reactive observation to proactive prediction of how information moves through our connected world.

Key Foundational Work

Exploring user social behaviors in mobile social applications (2011) — contributed 1 concept elements to this direction.
Daily Deals: Prediction, Social Diffusion, and Reputational Ramifications (2011) — contributed 1 concept elements to this direction.
Dynamic Resource Management in Clouds: A Probabilistic Approach (2012) — contributed 1 concept elements to this direction.
History & citationVersion 1 · 2026-10-04

Current · Permanent link

View frozen version 1

Cite this direction

Latest version (a living page):

@misc{lacuna_epidemiological_diffusion_modeling_8f6dfe04,
  title = {epidemiological diffusion modeling},
  author = {{Lacuna}},
  howpublished = {Lacuna research direction},
  year = {2026},
  url = {https://lacuna.tiptreesystems.com/direction/epidemiological-diffusion-modeling/rdr_8f6dfe04518c438084d4e8fd5c970135},
  note = {Latest version of a living research direction; cite a version for a fixed text. Accessed 2026-11-20},
}

This exact version (version 1, frozen text):

@misc{lacuna_epidemiological_diffusion_modeling_8f6dfe04_v1,
  title = {epidemiological diffusion modeling},
  author = {{Lacuna}},
  howpublished = {Lacuna research direction},
  year = {2026},
  url = {https://lacuna.tiptreesystems.com/direction/epidemiological-diffusion-modeling/rdr_8f6dfe04518c438084d4e8fd5c970135/v/1},
  note = {Version 1, frozen 2026-10-04. Accessed 2026-11-20},
}
About

These directions overlap with this one through shared papers.

See something wrong? Report a mistake