Quantifying Transmission Heterogeneity Using Both Pathogen Phylogenies and Incidence Time Series

Area of Science:

• Epidemiology
• Mathematical Biology
• Phylogenetics

Background:

• Individual-level transmissibility variation, quantified by the dispersion parameter k, impacts epidemic predictability and control.
• Estimating k from incidence time series alone is often insufficient.
• Phylogenetic analysis offers a complementary data source for estimating epidemiological parameters.

Purpose of the Study:

• To develop and validate an inference framework combining incidence data and pathogen phylogeny for estimating transmission heterogeneity (k) and other epidemiological parameters.
• To assess the impact of accurate k estimation on other model parameters, including the reproductive number.
• To evaluate the framework's performance under phylogenetic uncertainty and apply it to a real-world poliovirus outbreak.

Main Methods:

• Developed a particle Markov Chain Monte Carlo (pPMC) inference framework.
• Integrated incidence time series and pathogen phylogeny data.
• Employed a modified compartmental transmission model incorporating the dispersion parameter k.
• Utilized posterior distribution sampling for phylogenetic uncertainty.

Main Results:

• Combining epidemiological and phylogenetic data yielded more accurate and less biased reproductive number estimates.
• Pathogen phylogeny alone provided the most accurate estimation of k.
• Accurate k estimation was essential for unbiased reproductive number estimates but did not impact reporting probability or epidemic start date accuracy.
• Inference remained feasible despite significant phylogenetic uncertainty.

Conclusions:

• The developed pPMC framework effectively integrates epidemiological and phylogenetic data for robust parameter estimation.
• Accurate quantification of transmission heterogeneity (k) via phylogenetic analysis is vital for reliable epidemic modeling and control strategies.
• Phylogenetic data significantly enhances the accuracy and precision of epidemiological parameter estimates, even with substantial phylogenetic uncertainty.