In COVID-19-Modelling/BassSIR: Infectious disease modelling with Bass-SIR model

knitr::opts_chunk$set(echo = TRUE)

Model fitting

Under the structure of Bass-SIR model, we provide three models in the family for fitting

• Bass-SIR Bass-SIR model: $\frac{dI}{dt} = (\kappa + \beta I)\frac{m-I-A}{m} - \gamma I$
• SIR ordinary SIR model, which fixed $\kappa$ to zero
• Growth logistic growth model, which fixed $\beta$ to zero

Load the package as a starter

library(BassSIR)

Input data

We use the data of COVID-19 in Hubei as an example

cases <- as_bass_data(n_covid19$Hubei, id = "Hubei")

Fitting and summarising

est1 <- BassSIR::fit(cases, r_rec = 1/22.2, r_death = 1/22.3, type = "BassSIR")
est2  <- BassSIR::fit(cases, r_rec = 1/22.2, r_death = 1/22.3, type = "SIR")
est3  <- BassSIR::fit(cases, r_rec = 1/22.2, r_death = 1/22.3, type = "Growth")

summary(est1)

summary(est2)

summary(est3)

Model comparison

• DIC: Deviance Information Criteria, lower is better
• Deviance: narrower is more stable, lower is better
• Bayes factor: higher is better

compare_models(BassSIR = est1, SIR = est2, Growth = est3)

COVID-19-Modelling/BassSIR documentation built on April 20, 2020, 1:57 a.m.