Opportunities and Challenges in Developing COVID-19 Simulation Models: Lessons from Six Funded Projects

Abstract

The COVID-19 pandemic showed us the importance of modeling and forecasting efforts to guide decision makers. However, a year into the COVID-19 pandemic, the computational science literature lacks a proper internal exploration of the modeling journey of researchers around the world, including how they responded to the shared challenges our community faced such as data limitations, model fitting and working with public stakeholders. The current paper is a detailed examination of the internal processes of six research teams, which were funded in several countries to model COVID-19. Each team was asked to reflect on the research question and how they solved their respective modeling challenges, as well as how, looking back, they would do things differently.

A Scoping Review of Three Dimensions for Long-Term COVID-19 Vaccination Models: Hybrid Immunity, Individual Drivers of Vaccinal Choice, and Human Errors

“…Indeed, cognitive sciences show that humans only capture a small portion of the available information [ 100 ], sometimes making errors in storing this information, and ultimately interpreting it based on their own heterogeneous beliefs [ 101 ]. Prior works in modeling COVID-19 [ 28 , 29 , 33 ] and perceptual uncertainties in ABMs [ 102 ] have already accounted for some of these aspects when creating cognitive social simulations [ 103 ] that reflect the imperfect, heterogeneous decisions that individuals make on vaccination. While some works introduce the notion of imperfect decisions in COVID-19 by accounting for uncertainty about individual health states [ 104 ], research shows that there are at least three sources of individual errors to reflect that individuals use social information sub-optimally [ 105 ] when observing their peers (with respect to infection, vaccine choices, or death).…”

“…This ability to implement heterogeneity is indeed essential for COVID-19 since it is found in individual risk factors (e.g., age, hypertension, diabetes), contact patterns (e.g., social networks in the community or work settings), behaviors (e.g., willingness to be vaccinated), and spatial aspects (e.g., access to healthcare) [ 24 , 25 , 26 ]. Although ABMs for COVID-19 have been extremely varied in purpose and design [ 27 ], their design broadly followed three stages in the pandemic [ 28 ]: ABMs started with a handful of stages (e.g., susceptible, exposed, infected, recovered) and examined which non-pharmaceutical interventions would have the strongest effect [ 29 , 30 , 31 ], then ABMs were created to support vaccine-related studies (e.g., who should be vaccinated, where to place the centers and how to staff them) [ 32 , 33 , 34 ] and lastly, the current wave of studies where repeated boosters account for waning immunity [ 35 , 36 ].…”

Negotiating a Future that is not like the Past

“…Whilst not tackling those exact questions, ABM has been used during the Covid pandemic to model consequences of particular policy interventions (Ghorbani et al, 2020;Giabbanelli et al, 2020;Lorig et al, 2021). One example of this approach is JuSt-Social (Badham et al, 2021), an agent-based model developed to assist planners in North East England to understand the potential impact of COVID-19 on health and social service demand, particularly hospital bed requirements.…”

BESSIE: A Behavior and Epidemic Simulator for Use With Synthetic Populations

“…In the agent-based modeling and simulation community, many models and platforms have been developed, with goals including understanding the spread of the epidemic, forecasting, guiding policy-making, and evaluating counterfactuals [23,10]. While different goals may require different kinds of models, doing a detailed analysis of the effects of NPIs requires two main ingredients or components: adequate detail based on real data [24] and true agency (adaptive coupling with the environment, normativity, etc.…”