Chair:

Dr. Jason E. Shoemaker

Department of Chemical and Petroleum Engineering, University of Pittsburgh 

Committee Member:

Dr. Christopher R. Wilmer

Department of Chemical and Petroleum Engineering, University of Pittsburgh

Committee Member:

Dr. Ipsita Banerjee

Department of Chemical and Petroleum Engineering, University of Pittsburgh

Committee Member:

Dr. James R. Faeder

Department of Computational and Systems Biology, University of Pittsburgh

Title: Mathematical Modeling and Machine Learning Guided Optimization to Characterize Immunoregulation during Respiratory Infection

Abstract: Respiratory viruses present major public health challenges, as evidenced by seasonal influenza’s 290,000 – 650,000 global deaths annually, while the novel Severe Acute Respiratory Coronavirus 2 (SARS-CoV-2) has caused 6.28 million deaths worldwide. These viruses invoke excessive immune responses; however, the kinetics that regulate inflammatory responses during infection remain unresolved. Understanding the dynamics of the innate immune response and its manifestations at the cell and tissue levels is vital to understanding the mechanisms of immunopathology and to developing strain-independent treatments. Computational models of the innate immune response to respiratory infections are designed to provide greater insights into the regulation of the immune system, which will likely provide insights into clinical treatments and the pathological understandings of the disease. Efforts to develop these models have greatly increased in recent years as the use of RNA and protein level data have become widely available in public repositories.

We developed a suite of mathematical immune system models (Aims 1 – 3) and a methodology to find the minimum number of independently estimated parameters for a model to describe multiple cohorts or data sources, while all other parameters remain shared for all cohorts (Aim 4). Aim 1 incorporates viral replication, cell death, interferon stimulated genes’ effects on viral replication, and demonstrates that RIG-I antagonism significantly alters cytokine signaling trajectory. Aim 2’s model is a novel, spatialized, multicellular representation of RNA virus infection and type-I interferon-mediated antiviral response. The model suggests that selective degradation of extracellular virus or stabilization of interferon involved in paracrine signaling may improve host infection outcomes. Aim 3 compares low-pathogenic H1N1 and high-pathogenic H5N1 influenza virus infections, suggesting that the production rate of interferon is the major driver of strain-specific immune responses. Aim 4 details an unbiased method for determining the minimum number of parameters which allow a model to describe data from different experimental conditions (virus strain, treatment group, cohort), using an extension of Aim 3.

A greater understanding of the contributors to strain-specific immunodynamics can be utilized in future efforts aimed at treatment development to improve clinical outcomes of high-pathogenic viral strains. As model kinetics are host cell specific and not virus specific, the model presented provides an important step to modeling the intracellular immune dynamics of many RNA viruses, including the viruses responsible for SARS, Middle East Respiratory Syndrome (MERS), and COVID-19.