The Use Of Simulated Data In Analysing Parameter Identifiability For In-Host Compartmental Models Of mRNA Vaccine Immune Responses: An Application To Improving Sampling Of Physical Patient Data

Abstract

In–host mathematical models have become common place in mathematical epidemiology and immunology for the analysis of disease spread, treatment, and prevention. This theory has been extensively applied to modelling the immune responses of patients following vaccine administration by fitting data attained from clinical trials. Often such data is sparse and does not contain observations of all relevant compartments which can impact parameter identifiability. We consider an analysis of parameter identifiability of an in–host vaccine immune response model for LNP mRNA vaccines using simulated data. We employ the use of the software DAISY, for differential algebra analysis, and Monolix, for NMLE modelling that estimates model parameter values through fitting the simulated data. The results indicate that prior knowledge of T–cell and interleukin dynamics allowing for the fixing of parameters on a population level can yield accurate parameter estimates even when a limited number of compartments are observed.