Inherently high uncertainty in predicting the time evolution of epidemics
- Author:
Seung-Nam PARK
1
;
Hyong-Ha KIM
;
Kyoung Beom LEE
Author Information
- Publication Type:Original Article
- From:Epidemiology and Health 2021;43(1):e2021014-
- CountryRepublic of Korea
- Language:English
-
Abstract:
OBJECTIVES:Amid the spread of coronavirus disease 2019 (COVID-19), with its high infectivity, we have relied on mathematical models to predict the temporal evolution of the disease. This paper aims to show that, due to active behavioral changes of individuals and the inherent nature of infectious diseases, it is complicated and challenging to predict the temporal evolution of epidemics.
METHODS:A modified susceptible-exposed-infectious-hospitalized-removed (SEIHR) compartment model with a discrete feedback-controlled transmission rate was proposed to incorporate individuals’ behavioral changes into the model. To figure out relative uncertainties in the infection peak time and the fraction of the infected population at the peak, a deterministic method and 2 stochastic methods were applied.
RESULTS:A relatively small behavioral change of individuals with a feedback constant of 0.02 in the modified SEIHR model resulted in a peak time delay of up to 50% using the deterministic method. Incorporating stochastic methods into the modified model with a feedback constant of 0.04 suggested that the relative random uncertainty of the maximum fraction of infections and that of the peak time for a population of 1 million reached 29% and 9%, respectively. Even without feedback, the relative uncertainty of the peak time increased by up to 20% for a population of 100,000.
CONCLUSIONS:It is shown that uncertainty originates from stochastic properties of infections. Without a proper selection of the evolution scenario, active behavioral changes of individuals could serve as an additional source of uncertainty.
