A data simulation method to optimize a mechanistic dose-response model for viral loads of hepatitis A

Abstract

Driven by the quantitative estimate of risk via the dose-response models, quantitative microbial risk assessment has been used successfully for public health interventions. The dose-response models are derived starting from an average exposed dose of infectious particles, this dictates the dose data units required. Then dose-response data from animal model experiments are used to optimize these mechanistic dose-response models. For hepatitis A (Hep-A), the only available dose-response data use grams of feces for dose units. Therefore, to develop a dose-response model for Hep-A a method of converting these doses in grams of feces into infectious particles, while accounting for the uncertainty of this conversion is needed. This research develops a method to couple data simulation with the likelihood estimation method for model optimization to accomplish this. This adapted method uses data simulation to model the doses as viruses while accounting for the within-group variability of this simulation. Then these simulated doses, coupled with the original dose-response data, are used to optimize the mechanistic dose-response models. This method results in a more computationally rigorous means of modeling these types of dose-response data. The resulting dose-response model for Hep-A is also more appropriate to use than the current option for Hep-A risk models.