CALMS: Modelling the long-term health and economic impact of Covid-19 using agent-based simulation

Abstract

We present our agent-based CoronAvirus Lifelong Modelling and Simulation (CALMS) model that aspires to predict the lifelong impacts of Covid-19 on the health and economy of a population. CALMS considers individual characteristics as well as comorbidities in calculating the risk of infection and severe disease. We conduct two sets of experiments aiming at demonstrating the validity and capabilities of CALMS. We run simulations retrospectively and validate the model outputs against hospitalisations, ICU admissions and fatalities in a UK population for the period between March and September 2020. We then run simulations for the lifetime of the cohort applying a variety of targeted intervention strategies and compare their effectiveness against the baseline scenario where no intervention is applied. Four scenarios are simulated with targeted vaccination programmes and periodic lockdowns. Vaccinations are targeted first at individuals based on their age and second at vulnerable individuals based on their health status. Periodic lockdowns, triggered by hospitalisations, are tested with and without vaccination programme in place. Our results demonstrate that periodic lockdowns achieve reductions in hospitalisations, ICU admissions and fatalities of 6-8% compared to the baseline scenario, with an associated intervention cost of £173 million per 1,000 people and targeted vaccination programmes achieve reductions in hospitalisations, ICU admissions and fatalities of 89-90%, compared to the baseline scenario, with an associated intervention cost of £51,924 per 1,000 people. We conclude that periodic lockdowns alone are ineffective at reducing health-related outputs over the long-term and that vaccination programmes which target only the clinically vulnerable are sufficient in providing healthcare protection for the population as a whole.

Mathematical models of COVID-19 vaccination in high-income countries: A systematic review

“…Following full text screening of 358 studies, 40 articles were included, having met our eligibility criteria. 21–60 A further six eligible studies were identified through screening reference lists of included studies, 61–66 and one additional study was discovered through searching for papers describing the European COVID-19 Scenario Hub models, 67 giving a total of 47 studies included in our review. Out of the 47 included studies, there were 40 unique models; five models were used in more than one paper.…”

Large-Scale Parallelization of Human Migration Simulation

“…Examples of these include the propagation of knowledge across communities and the spread of infectious diseases on a local, individualized level. To specifically address the latter challenge, we instead developed a different simulation kernel (the Flu And Coronavirus Simulator [33], [35]) which is better suited to efficiently calculate the outcomes of large numbers of interpersonal interactions.…”

Proceedings of SW21 The OR Society Simulation Workshop

“…Three simulation models were used in the workshop that model different aspects of the COVID-19 pandemic. These include the dynamiC Hospital wArd Management (CHARM) Discrete-Event Simulation (DES) for hospital bed capacity management (Anagnostou et al, 2022), the Flu and Coronavirus Simulation (FACS) Agent-Based Simulation (ABS) that models the geospatial spread of the disease and the impact of preventive measures (Anagnostou et al, 2022), and the CoronAvirus Lifelong Modelling and Simulation (CALMS) ABS that considers long-term policy intervention effectiveness (Mintram et al, 2022).…”