Speaker
Description
Infectious disease models are common at both the cellular scale to describe invasion probabilities and the tissue scale to describe pathogen replication dynamics once an infection has been established. We develop Bayesian methods that link the infection establishment with mechanistic target cell ODE models of viral replication. First, given longitudinal measurements of hosts’ viral load, we estimate ODE parameters using Approximate Bayesian Computation. Next, we develop a statistical model where infection status (positive or negative) is modeled as a latent variable, and observations of the viral load are described assuming a censored lognormal distribution. Then, the statistical model is fit using a Hybrid Expectation-Maximization algorithm. We apply the model to influenza human-subject challenge studies where i) the time of infection is known and ii) multiple baseline immune system measures are collected and discover protective thresholds for immunity against infection.
