MS06 - MEPI-06 Part 2 of 3

Recent Advances in Dynamics of Human Behavior and Epidemics (Part 2)

Thursday, July 17 at 10:20am

SMB2025 Follow

Note: this minisymposia has multiple sessions. The other sessions are: Part-1 and Part-3.

Abba Gumel (University of Maryland), Alex Safsten, Alice Oveson (both University of Maryland)

Description:

The recent COVID-19 pandemic has highlighted the critical role human behavior plays in the dynamics and control of infectious diseases. The behavior changes with respect to the adherence or lack thereof to public health intervention and mitigation measures during this pandemic were triggered by factors such as the unprecedented burden of the disease, the epidemic of disease-related mis(dis)information, fear, polarization, peer influence, poor quality and inconsistency in public health messaging, etc. Specifically, epidemiological models that do not explicitly account for these behavioral changes were seen to generally fail to capture the correct trajectory and burden of the pandemic (thereby not being able to make realistic or accurate forecasts). This minisymposium brings together an interdisciplinary team of researchers to discuss and share ideas on the recent advances in designing, validating, and analyzing mathematical models that explicitly incorporate human behavior and socio-economic factors, and use these models to contribute to public health policy for controlling and mitigating the spread and burden of the disease. Some of the topics to be addressed include metrics of human behavior changes, the role of heterogeneity in compliance to public health intervention and mitigation measures, the influence of social networks, the impact of mis(dis)information, and risk perception.

Mallory Harris

University of Maryland

"Risk (Mis)estimation and Population Heterogeneity Shape Infectious Disease Dynamics"

Models of human behaviour during infectious disease outbreaks often assume that people perfectly assess the risks associated with infection and become more cautious when risk is high. However, prior work showed that people tended to misestimate the risk of Covid-19 exposure at events of different sizes (Sinclair et al 2021, PNAS). The effects of event risk estimation have not been studied at population level, a critical gap given potential for nonlinear and emergent dynamics in infectious disease systems. Here, we build an agent-based model to capture feedback between infectious disease dynamics, risk perception, and behavior in the context of event attendance. At each time step, individuals decide whether to attend an event based on their assessed exposure risk, a function of event size and prevalence calibrated to actual risk assessments collected from 11,169 individuals across the United States between September 2021 and August 2022 (Sinclair et al 2023, PLoS One). We show that risk misestimation substantially worsens epidemic burden compared to what it would be if people estimated risk perfectly. Behavioural interventions to improve risk estimation reduce but do not completely eliminate this effect. We also compare strategies for deploying behavioural interventions across a heterogeneous population where certain subgroups are more likely to underestimate risk. This work underscores the importance of considering risk misestimation in mathematical models of infectious diseases and demonstrates benefits of behavioural interventions to improve individual decision-making and reduce disease transmission. Joint work with Shu Yuan Shi and Joshua Weitz.

Christian Parkinson

Michigan State University

"Optimal Control of a Reaction-Diffusion Epidemic Model with Noncompliance"

We consider an optimal distributed control problem for a reaction-diffusion-based SIR epidemic model with human behavioral effects. We develop a model wherein non-pharmaceutical intervention methods are implemented, but a portion of the population does not comply with them, and this noncompliance affects the spread of the disease. Drawing from social contagion theory, our model allows for the spread of noncompliance parallel to the spread of the disease. Control variables affect the infection rate among the compliant population, the rate of spread of noncompliance, and the rate at which non-compliant individuals return to a compliant state. We prove the existence of global-in-time solutions for fixed controls and study the regularity properties of the resulting control-to-state map. We establish the existence of optimal controls for a fairly general class of objective functions and present a first-order stationary system which is necessary for optimality. Finally, we present simulations with various parameters values to demonstrate the behavior of the model.

Zitao He

University of Waterloo

"From Sentiment to Spread: Homophily and Early Warnings in Epidemic Dynamics"

Understanding the interplay between social activities and disease dynamics is crucial for effective public health interventions. While many coupled behavior-disease models assume homogeneous populations, real-world social structure is often heterogeneous. In this talk, we present a model that divides the population into social media users and non-users to investigate the impact of homophily (the tendency for individuals to associate with similar others) and online events on disease dynamics. We find that homophily slows down the spread of vaccinating strategies, pushing the system closer to a tipping point where vaccine uptake collapses and an endemic equilibrium emerges. Online events also play an important role, with early social media discussions acting as warning signs of upcoming outbreaks. Building on these insights, we also discuss a data-driven approach that uses deep learning to detect early warning signals from vaccine-related social media time series. Specifically, we train LSTM and ResNet classifiers on simulated data from a stochastic behavior-disease model with additive Lévy noise, capturing heavy-tailed real-world fluctuations. These classifiers consistently outperform conventional indicators such as variance and lag-1 autocorrelation, offering clearer and more interpretable signals. Together, these studies underscore the importance of incorporating social structure and real-time data in predictive models for proactive public health response.

Alice Oveson

University of Maryland

"Modeling Racial and Age-Structured Transmission Dynamics with Empirical Contact Data"

I present a compartmental infectious disease model structured by both race and age, incorporating empirically derived contact matrices to represent heterogeneity in social behavior. The model captures interactions across twelve demographic subgroups and enables the study of how behavioral mixing patterns shape disease transmission. While the inclusion of contact data explains a substantial portion of variation in group-level transmission dynamics, our results indicate that racial disparities persist beyond what can be attributed to behavioral contact patterns alone. This suggests the influence of unmeasured structural factors, such as differential susceptibility, healthcare access, or baseline risk. Our approach highlights the utility of structured modeling frameworks for uncovering the multi-layered mechanisms underlying population-level disparities in disease burden.

#SMB2025 Follow

Annual Meeting for the Society for Mathematical Biology, 2025.