Stochastic Spillover Model Identifies Top Parameter Uncertainties in Americas Yellow Fever Variation

Uncertainties in only a handful of parameters greatly affect variation in cumulative disease incidence within an illustrative stochastic model of yellow fever virus spillover in the Americas. Specifically, the modeling analysis shows that uncertainty in patch connectivity and uncertainty in non-human primate latency have the greatest impact on modeled variation in cumulative disease incidence of yellow fever virus across patches. Researchers obtained these model findings by applying tools from the field of uncertainty quantification to separate two channels of noise in stochastic analyses of infectious disease emergence.

In the study, the authors show how tools from the field of uncertainty quantification allow rigorous statistical inferences about processes important for disease emergence. They illustrate these techniques with a stochastic model of yellow fever virus spillover in the Americas, examining a virus that has seen rapid re-emergence among multiple hosts and vectors in South America over the last decade. Because this yellow fever analysis evaluates an illustrative spillover model, its conclusions specifically describe how parameter uncertainties affect modeled cumulative incidence variation across patches.

The modeling framework addresses limitations that emerging infectious diseases pose as a persistent public health threat to deterministic, mechanistic modeling approaches. Because outbreaks initially start with a low number of infected hosts, their dynamics are highly stochastic. Consequently, traditional deterministic methods, such as ordinary differential equations, are unable to qualitatively or quantitatively capture transmission dynamics when an outbreak begins with few infected hosts.

To represent stochastic outbreak dynamics, researchers instead use stochastic models, such as Markov chain models, though these models present challenges of their own. Inferring a quantity of interest with stochastic models often requires sampling the model’s distribution many times over, introducing an additional source of noise that makes inference more difficult and computationally expensive. By efficiently separating these two channels of noise, the uncertainty quantification approach pinpoints uncertainties in patch connectivity and non-human primate latency as having the greatest impact on modeled cumulative incidence variation across patches.

Subject of Research: Uncertainty quantification for stochastic infectious disease models illustrated with yellow fever virus spillover in the Americas

Article Title: Global sensitivity analysis for stochastic models of infectious diseases: Re-emergence of yellow fever virus in the Americas as a case study

Article References: Kornetzke, N., & Wearing, H. J. (2026). Global sensitivity analysis for stochastic models of infectious diseases: Re-emergence of yellow fever virus in the Americas as a case study. PLOS Computational Biology, 22(11), e1013952. https://doi.org/10.1371/journal.pcbi.1013952

Image Credits: AI Generated

DOI: 10.1371/journal.pcbi.1013952

Keywords: uncertainty quantification, stochastic models, yellow fever virus, yellow fever spillover, emerging infectious diseases, Markov chain models, patch connectivity, non-human primate latency, Global, sensitivity, analysis, stochastic