Abstract: Vaccines are widely considered the most successful examples of medical therapeutics. Vaccines over the past few decades have been able to nearly eradicate a handful of lethal infections from the population. The cornerstone of successful vaccines is the ability to generate neutralizing antibodies against respective pathogens. Specifically newborn vaccines have proven to be the most successful in significantly reducing mortality from several otherwise fatal infections. Vaccines are an important source of immune protection but are only a piece of the immune protection puzzle. Additionally, all infants have some level of immune protection from antibodies transferred from the mother during pregnancy.

We aim to better understand the maternal-fetal antibody response with the hypothesis that it will provide novel insights into fundamental immune processes relating to infant protection from infection. Looking at pregnant women and infants we dissect the maternal-fetal antibody response to different pathogens (SARS-CoV-2 and CMV) with respect to vaccination and natural infection. Antibody profiles of natural infection are further explored in primary and latent CMV infection. Leveraging this experimental data machine learning and deep learning are applied to further dissect the antibody immune response. Specifically, the kinetics of the antibody response are explored over time and predictions are made on infection status. Deep learning is leveraged to go further and explore viral transmission dynamics from mother to infant. We successfully predict if a mother will transmit CMV to her infant where no diagnostic method is currently available. Understanding different aspects of the antibody response at the maternal fetal interface can provide new insights that may be utilized for new vaccine and therapeutic development against infectious diseases that currently have limited therapeutic options.

Thesis Committee: Margaret Ackerman (Chair), Jiwon Lee, Saeed Hassanpour, Ali Ellebedy