Forecasting of COVID-19 spread in Poland as part of health needs mapping

Abstract

Abstract The development of a robust model for COVID-19 forecasting was a crucial task undertaken by the Ministry of Health of the Republic of Poland in response to the pandemic. High-quality forecasting rendered the inclusion of COVID-19 modelling in health needs assessment indispensable at the regional as well as country level. The model served as a basis for data-driven policy-making, such as the introduction or lifting of non-pharmaceutical interventions and the management of hospital resource allocation. The proposed model draws upon the population-adjusted infection fatality rates, vaccine-induced immunity, reported test positivity rates, and COVID-19-related mortality to infer the dark figure of cases. The estimates of the true number of infections lay the foundations for the computation of variant-dependent effective reproduction numbers, which constitute the workhorse of the model. The algorithm utilises the Bayesian prediction of those metrics along with the estimation of hybrid immunity levels to arrive at a short-term forecast of the course of the pandemic. The implemented model yields reliable results throughout the outbreak, allowing for the assessment of the strain of consecutive epidemic waves on the healthcare system. Key messages • Forecasting of COVID-19 spread is vital for health needs assessment and health crisis management policy-making. • Short-term forecasting of COVID-19 spread using effective reproduction numbers yields high-quality results.