Robust and Intuitive Model for COVID-19 Epidemic in Slovenia
Keywords:Epidemic, COVID-19, Modelling, SEIR, Reproduction number, Public health interventions
The main goal of epidemic modelling is to support the epidemic management through forecasts and analyses of past developments. With this in mind a robust and intuitive SEIR (Susceptible, Exposed, Infectious, Recovered) type model has been developed, applied and verified during the multiple waves of the COVID-19 epidemics in Slovenia since March 2020. The model parameters were based on the general characteristics of the COVID-19 disease reported globally for the entire planet and refined with the aggregate data available mostly on a daily basis in Slovenia, as for example the number of confirmed cases, hospitalized patients, hospitalized patients in intensive care units and deceased. The Slovenian aggregate data was also used to estimate the degree of immunisation due to past infections and vaccination, which reduces the number of susceptible persons for the disease.
Examples of the model application are presented to illustrate its robustness and intuitiveness in both the forecasts and analyses of past developments. The analyses of past developments provided specific estimates of modelling parameters for Slovenia and quantified the effects of pharmaceutical and non-pharmaceutical interventions and various events on the development of the epidemics as measured through the reproduction number R. This empirically obtained information was then applied in the forecasts. Accurate forecasts are a great support for decision makers and for hospitals to plan appropriate actions in advance. The inherent uncertainties in the model and data were quantified through intuitive sensitivity analyses represented as different scenarios. The observed accuracy of the forecasts was impressively good also in demanding conditions, when various complex processes influencing the spread of the disease were going on in parallel. This demonstrates the robustness and relevance of the proposed model.
How to Cite
Copyright (c) 2022 The Authors
This work is licensed under a Creative Commons Attribution 4.0 International License.