The effect of time distribution shape on simulated epidemic models

TL;DR

This paper investigates how changing the shape of time distributions in epidemic models from exponential to gamma affects the simulation outcomes, revealing delays in disease spread with more realistic timing assumptions.

Contribution

It demonstrates the impact of using gamma distributions instead of exponential ones for latency and infectious periods in epidemic simulations, providing insights into more accurate modeling.

Findings

Delay in epidemic spread with gamma distributions

More realistic timing alters key simulation results

Explains the effect of distribution shape on epidemic dynamics

Abstract

By convention, and even more often, as an unintentional consequence of design, time distributions of latency and infectious durations in stochastic epidemic simulations are often exponential. The skewed distribtion typically leads to unrealistically short times. We examine the effects of altering the distribution latency and infectious times by comparing the key results after simulation with exponential and gamma distributions in a homogeneous mixing model aswell as a model with regional divisions connected by a travel intensity matrix. We show a delay in spread with more realistic latency times and offer an explanation of the effect.