Immune Network Modeling Predicts Specific Nasopharyngeal and Peripheral Immune Dysregulation in Otitis-Prone Children

Abstract

Acute otitis media (AOM) pathogenesis involves nasopharyngeal colonization by potential otopathogens and a viral co-infection. Stringently-defined otitis prone (sOP) children show characteristic patterns of immune dysfunction. We hypothesized that otitis proneness is largely a result of altered signaling between immune components that are otherwise competent, resulting in increased susceptibility to infection by bacterial otopathogens. To test this, we constructed a regulatory immune network model linking immune cells and signaling elements known to be involved in AOM and/or dysregulated in sOP children. The alignment of immune response mechanisms with data from in vivo and in vitro experimental observations produced 82 putative immune network models, each describing variants of immune regulatory networks consistent with available observations. Analysis of these models suggested that new measurements of serum levels of IL-4 and CXCL8 could refine competing models and resulted in the elimination of 38 of the models. Further analysis of the remaining 44 models suggested specific deviations in the predicted regulation of nasopharyngeal and peripheral immunity during response to AOM. Specifically, immune responses active in sOP children during AOM were characterized by early and constitutive activation of pro-inflammatory signaling in the nasopharynx and a Th2-and Treg-dominated profile in the periphery. We conclude that sOP children have altered regulation of key immune mediators during both health and pathogenesis. This altered regulation may be amenable to therapeutic intervention.

Two heads are better than one: current landscape of integrating QSP and machine learning

“…Semi-automated extraction of parameters from the literature has also been accomplished using natural language processing, however additional analysis is required to ensure the extracted data are applicable in a particular QSP setting. Together with a ML model checking framework [ 26 ], this approach has been applied to a combined of parameter selection and QSP model selection to model immune dysregulation in children prone to a specific viral infection [ 27 ].…”