Masking Behaviors in Epidemiological Networks with Cognitively-plausible Reinforcement Learning (preprint)

The COVID-19 pandemic highlighted the critical role of human behavior in influencing infectious disease transmission and the need for models capturing this complex dynamic. We present an agent-based model integrating an epidemiological simulation of disease spread with a cognitive architecture driving individual mask-wearing decisions. Agents decide whether to mask based on a utility function weighting factors like peer conformity, personal risk tolerance, and mask-wearing discomfort. By conducting experiments systematically varying behavioral model parameters and social network structures, we demonstrate how adaptive decision-making interacts with network connectivity patterns to impact population-level infection outcomes. The model provides a flexible computational framework for gaining insights into how behavioral interventions like mask mandates may differentially influence disease spread across communities with diverse social structures. Findings highlight the importance of integrating realistic human decision processes in epidemiological models to inform policy decisions during public health crises.

Should we Mitigate or Suppress the next Pandemic? Time-Horizons and Costs shape optimal Social Distancing Strategies (preprint)

The COVID-19 pandemic has called for swift action from local governments, which have instated Nonpharmaceutical Interventions (NPIs) to curb the spread of SARS-Cov-2. The quick and decisive decision to save lives through blunt instruments has raised questions about the conditions under which decision-makers should employ mitigation or suppression strategies to tackle the COVID-19 pandemic. More broadly, there are still debates over which set of strategies should be adopted to control different pandemics, and the lessons learned for SARS-Cov-2 may not apply to a new pathogen. While curbing SARS-Cov-2 required blunt instruments, it is unclear whether a less-transmissible and less-deadly emerging pathogen would justify the same response. This paper illuminates this question using a parsimonious transmission model by formulating the social distancing lives vs. livelihoods dilemma as a boundary value problem. In this setup, society balances the costs and benefits of social distancing contingent on the costs of reducing transmission relative to the burden imposed by the disease. To the best of our knowledge, our approach is distinct in the sense that strategies emerge from the problem structure rather than being imposed a priori . We find that the relative time-horizon of the pandemic (i.e., the time it takes to develop effective vaccines and treatments) and the relative cost of social distancing influence the choice of the optimal policy. Unsurprisingly, we find that the appropriate policy response depends on these two factors. We discuss the conditions under which each policy archetype (suppression vs. mitigation) appears to be the most appropriate.

Optimizing Social Distancing Policies: A Dynamic Programming Approach for Coupled High and Low Risk Populations (preprint)

Background: Decision makers may use social distancing to reduce transmission between risk groups in a pandemic scenario like Covid-19. However, it may result in both financial, mental, and social costs. Given these tradeoffs, it is unclear when and who needs to social distance over the course of a pandemic when policies are allowed to change dynamically over time and vary across different risk groups (e.g., older versus younger individuals face different Covid-19 risks). In this study, we examine the optimal time to implement social distancing to optimize social utility, using Covid-19 as an example. Methodology: We propose using a Markov decision process (MDP) model that incorporates transmission dynamics of an age-stratified SEIR compartmental model to identify the optimal social distancing policy for each risk group over time. We parameterize the model using population-based tracking data on Covid-19 within the US. We compare results of two cases: allowing the social distancing policy to vary only over time, or over both time and population (by risk group). To examine the robustness of our results, we perform sensitivity analysis on patient costs, transmission rates, clearance rates, mortality rates. Results: Our model framework can be used to effectively evaluate dynamic policies while disease transmission and progression occurs. When the policy cannot vary by subpopulation, the optimal policy is to implement social distancing for a limited duration at the beginning of the epidemic; when the policy can vary by subpopulation, our results suggest that some subgroups (older adults) may never need to socially distance. This result may occur because older adults occupy a relatively small proportion of the total population and have less contact with others even without social distancing. Conclusion: Our results show that the additional flexibility of allowing social distancing policies to vary over time and across the population can generate substantial utility gain even when only two patient risk groups are considered. MDP frameworks may help generate helpful insights for policymakers. Our results suggest that social distancing for high-contact but low-risk individuals (e.g., such as younger adults) may be more beneficial in some settings than doing so for low-contact but high-risk individuals (e.g., older adults).

Modeling the Impact of Social Distancing and Targeted Vaccination on the Spread of COVID-19 through a Real City-Scale Contact Network (preprint)

We use mobile device data to construct empirical interpersonal physical contact networks in the city of Portland, Oregon, both before and after social distancing measures were enacted during the COVID-19 pandemic. These networks reveal how social distancing measures and the public's reaction to the incipient pandemic affected the connectivity patterns within the city. We find that as the pandemic developed there was a substantial decrease in the number of individuals with many contacts. We further study the impact of these different network topologies on the spread of COVID-19 by simulating an SEIR epidemic model over these networks, and find that the reduced connectivity greatly suppressed the epidemic. We then investigate how the epidemic responds when part of the population is vaccinated, and we compare two vaccination distribution strategies, both with and without social distancing. Our main result is that the heavy-tailed degree distribution of the contact networks causes a targeted vaccination strategy that prioritizes high-contact individuals to reduce the number of cases far more effectively than a strategy that vaccinates individuals at random. Combining both targeted vaccination and social distancing leads to the greatest reduction in cases, and we also find that the marginal benefit of a targeted strategy as compared to a random strategy exceeds the marginal benefit of social distancing for reducing the number of cases. These results have important implications for ongoing vaccine distribution efforts worldwide.

Reopening California: Seeking Robust, Non-Dominated COVID-19 Exit Strategies (preprint)

Amid global scarcity of COVID-19 vaccines and the threat of new variant strains, California and other jurisdictions face the question of when and how to implement and relax COVID-19 Nonpharmaceutical Interventions (NPIs). While policymakers have attempted to balance the health and economic impacts of the pandemic, decentralized decision-making, deep uncertainty, and the lack of widespread use of comprehensive decision support methods can lead to the choice of fragile or inefficient strategies. This paper uses simulation models and the Robust Decision Making (RDM) approach to stress-test Californias reopening strategy and other alternatives over a wide range of futures. We find that plans which respond aggressively to initial outbreaks are required to robustly control the pandemic. Further, the best plans adapt to changing circumstances, lowering their stringent requirements to reopen over time or as more constituents are vaccinated. While we use California as an example, our results are particularly relevant for jurisdictions where vaccination roll-out has been slower.

Modeling COVID-19 Nonpharmaceutical Interventions: Exploring periodic NPI strategies (preprint)

In April 2020, we developed a COVID-19 transmission model used as part of RANDs web-based COVID-19 decision support tool that compares the effects of different nonphar-maceutical public health interventions (NPIs) on health and economic outcomes. An interdis-ciplinary approach informed the selection and use of multiple NPIs, combining quantitative modeling of the health/economic impacts of interventions with qualitative assessments of other important considerations (e.g., cost, ease of implementation, equity). We previously published a description of our approach as a RAND report describing how the epidemiological model, the economic model, and a systematic assessment of NPIs informed the web-tool. This paper provides further details of our model, describes extensions that we made to our model since April, presents sensitivity analyses, and analyzes periodic NPIs. Our findings suggest that there are opportunities to shape the tradeoffs between economic and health outcomes by carefully evaluating a more comprehensive range of reopening policies. We consider strategies that periodically switch between a base NPI level and a higher NPI level as our working example.