Calibrating COVID-19 susceptible-exposed-infected-removed models with time-varying effective contact rates

Abstract

We describe the population-based susceptible-exposed-infected-removed (SEIR) model developed by the Irish Epidemiological Modelling Advisory Group (IEMAG), which advises the Irish government on COVID-19 responses. The model assumes a time-varying effective contact rate (equivalently, a time-varying reproduction number) to model the effect of non-pharmaceutical interventions. A crucial technical challenge in applying such models is their accurate calibration to observed data, e.g. to the daily number of confirmed new cases, as the history of the disease strongly affects predictions of future scenarios. We demonstrate an approach based on inversion of the SEIR equations in conjunction with statistical modelling and spline-fitting of the data to produce a robust methodology for calibration of a wide class of models of this type. This article is part of the theme issue ‘Data science approaches to infectious disease surveillance’.

Inferring the effective reproductive number from deterministic and semi-deterministic compartmental models using incidence and mobility data

“…In the context of Ireland's COVID-19 epidemic, we derive similar < 0 estimates from the three DGPs (DGP1: 95% CI[4.5-6.9], DGP2: 95% CI [4.4-6.8], DGP3: 95% CI[5.8-7.0]). These estimates are in close agreement with a previous modelling study on the COVID-19 pandemic in Ireland [35], albeit well above the initially reported < 0 = 2.2 value from Wuhan [60]; a value that has been adopted as the reference point by the World Health Organization and other research groups [15,61]. Other streams of research, however, argue that the initial estimate was low [62], and instead, advocate for higher values (4.5 [62]; 4.7-6.6 [63]).…”

“…system of ordinary differential equations dx dt ¼ f ðxÞ is Markovian. Here, we adopt the SEI3R profile [15,35], an extension of the SEIR framework. Under this profile, we stratify individuals as susceptible (S t ), exposed (E t ), infectious, and recovered (R t ).…”

A COVID-19 model incorporating variants, vaccination, waning immunity, and population behavior

“…Gonzalez-Parra et al 34 used a modified compartment model that incorporated latent, infective symptomatic, and symptomatic individuals to take into account VOC such as Alpha variant (B.1.1.7) to predict prevalence, hospitalizations, and deaths. However, despite numerous useful studies [35][36][37][38][39] , there remains a gap in modeling the coupled effects of population behavioral changes, waning immunity, vaccine roll-out strategies, and new VOCs on the progression of the pandemic.…”

Attention based parameter estimation and states forecasting of COVID-19 pandemic using modified SIQRD Model

“…The parameters of these mathematical models capture the global behavior of the trajectories of epidemic numbers, which may be inferred or interpreted as the behavior of the population in terms of mobility [20] , as well as transmitability of virus [31] that may lead to the implementation of measures [21] like social distancing, quarantine management, mandatary mask wearing and closing of public and private places. In literature, these parameters are either assumed, or estimated through optimization techniques from the domain of nature-inspired algorithms [23] , Bayesian algorithms [32] and gradient-based algorithms [33] , in application of fitting the real world data over differential system [34] , basis functions [35] , spline fitting [36] , Kalman filter [37] , neural nets [38] and even recurrent nets [33] . Furthermore, stochastic [31] or smooth [36] perturbations in the parameters of these models have been studied that lead to the better fitting of the epidemic models over real-world data.…”