Inferring epidemic containment probability from non-pharmaceutical
interventions using a Gillespie-based epidemic model
Abstract
The Gillespie algorithm, which is a stochastic numerical simulation of
continuous-time Markovian processes, has been proposed for simulating
epidemic dynamics. In the present study, using the Gillespie-based
epidemic model, we focused on each single trajectory by the stochastic
simulation to infer the probability of controlling an epidemic by
non-pharmaceutical interventions (NPIs). The single trajectory analysis
by the stochastic simulation suggested that a few infected people
sometimes dissipated spontaneously without spreading of infection. The
outbreak probability was affected by basic reproductive number but not
by infectious duration and susceptible population size. A comparative
analysis suggested that the mean trajectory by the stochastic simulation
has equivalent dynamics to a conventional deterministic model in terms
of epidemic forecasting. The probability of outbreak containment by NPIs
was inferred by trajectories derived from 1000 Monte Carlo simulation
trials using model parameters assuming COVID-19 epidemic. The
model-based analysis indicated that complete containment of the disease
could be achieved by short-duration NPIs if performed early after the
import of infected individuals. Under the correctness of the model
assumptions, analysis of each trajectory by Gillespie-based stochastic
model would provide a unique and valuable output such as the
probabilities of outbreak containment by NPIs.