Enhancing Accuracy in COVID-19 Hospital Admission Estimates: Integrating Chronic Conditions into an Agent-Based Model

Keywords: COVID-19, hospitalization, chronic conditions, agent-based modeling
Abstract Type: Paper Abstract

Abstract

Individuals with chronic conditions, such as asthma, chronic obstructive pulmonary disease (COPD), diabetes, hypertension, and obesity, are at high risk for poor health outcomes from COVID-19. Accurately accounting for the heightened risk of hospitalization for individuals with these conditions is crucial for healthcare emergency preparedness. An agent-based model called the Simulator for Infectious Disease Dynamics in North Carolina (SIDD-NC) has been developed to simulate the actions and interactions of autonomous agents during the COVID-19 pandemic. However, a limitation of SIDD-NC is its sole reliance on age-based probabilities to forecast hospital admission estimates. Thus, we employ a novel approach of assigning conditions to agents in a simulated North Carolina population, for which we consider differences in condition prevalence between both age and race groups. We then incorporate condition-specific hospitalization risk into SIDD-NC. The model outputs will be evaluated against ground truth data from the North Carolina Department of Health and Human Services (NCDHHS) for the number of new daily COVID-19 hospital admissions during the first year of the pandemic. Specifically, using both Spearman’s rank correlation and Root Mean Square Error (RMSE), the model outputs will be assessed both spatially (state and region) and temporally. Public health decision-makers and health care professionals may use the model to forecast hospital admissions, aiding hospitals in preparing staff and resources for future COVID-19 surges and novel pandemics.