Skip to main content

Table 1 Overview of models studied

From: A comparison of five epidemiological models for transmission of SARS-CoV-2 in India

Name of model Comments Input(s) and output(s) Parameter(s) and estimation
Baseline (Bhardwaj, R. 2020) [17] Curve-fitting model.
Cumulative number of infected cases modeled as exponential process, with growth rate λ.
Daily time series of number of infected individuals from T0 till T11 (as input) and from T1 to T22 (as output). Time varying growth rate of infection is estimated from input and modeled using least-squares regression. Estimation involves implementing MCMC3 methods for a Bayesian framework.
eSIR (Wang, L. et al., 2020) [23,24,25] Extension of the standard SIR2 compartmental model. Daily time series data on proportion of infected and recovered individuals from T0 till T11 (as input) and from T1 to T22 along with posterior distribution of parameters and prevalence values of the three compartments in the model (as output). β and γ control transmission and removal rates respectively. λ and κ control variability of observed and latent processes respectively. Estimation involves implementing MCMC3 methods for a hierarchical Bayesian framework.
SAPHIRE (Hao, X. et al., 2020) [13] Extension of the standard SEIR2 compartmental model. Daily time series data from T0 till T11 on count of infected individuals (as input) and count of infected and removed individuals from T1 to T22 along with posterior distributions of parameters (as output). Unreported cases are also presented. See Section 2.1.c for details on parameters. Estimation involves implementing MCMC3 methods for a Bayesian framework.
SEIR-fansy (Bhaduri, R., Kundu, R. et al., 2020) [18, 26] Another extension of standard SEIR2, accounting for the possible effect of misclassifications due to imperfect testing. Daily time series data from T0 till T11 on proportion of dead, infected and recovered individuals (as input) and from T1 to T22 along with posterior distributions of parameters and prevalence values of compartments in the model (as output). Unreported cases and deaths are also projected. See Supplementary Table S1 for details on parameters. Estimation involves implementing MCMC3 methods for a hierarchical Bayesian framework.
ICM (Flaxman et.al., 2020) [7] Renewal equation used to model infections as a latent process. Deaths are linked to infections via a survival distribution. Accounts for changes in behavior and various governmental policies enacted. Daily time series data from T0 till T11 on count of dead individuals (as input) and from T1 to T2 (as output). Posterior over infections, deaths and various parameters.
Infections include both symptomatic and asymptomatic ones.
See Section 2.1.e for details on parameters.
Estimation is done via HMC4 using STAN.
  1. 1 T0: time of crossing 50 confirmed cases – March 12, 2020. T1: October 15, 2020. T2: December 31, 2020
  2. 2 S(E)IR susceptible-(exposed)-infected-removed
  3. 3 MCMC Markov chain-Monte Carlo
  4. 4 Hamiltonian Monte Carlo