Mathematical and Statistical Modeling of Epidemiology of Asymptomatic COVID-19 Infections and Vaccine Design Strategies

Abstract

A defining feature of the COVID-19 pandemic has been the high prevalence of asymptomatic SARS-CoV-2 infections – laboratory-confirmed cases where individuals never develop symptoms. Accurately estimating the proportion of such infections (ψ) is crucial for refining transmission models, designing equitable public health strategies, and deepening our understanding of host-pathogen interactions. Yet, conventional approaches are hampered by nonspecific symptom definitions and diagnostic test imperfections, leading to substantial underestimation of asymptomatic cases.

This thesis addresses these challenges through a quantitative framework that integrates epidemiological data analysis with mechanistic modeling of immune responses. I begin by developing a mathematical formalism that corrects for both test inaccuracies and symptom misclassification in seroprevalence surveys. Applying this framework to 50 serosurveys across 29 countries – spanning approximately 800,000 individuals – I show that the true prevalence of asymptomatic infection is significantly higher than reported (median ψ ≈ 60% vs. reported ψc ≈ 40%, P = 4×10−7). The correction primarily arises from symptom overlap, which explains nearly 85% of the observed discrepancy, and leads to improved consistency within countries.

To explore global patterns in asymptomatic infection, I conduct a review and meta-analysis of 26 nationally representative serosurveys conducted prior to mass vaccination. The analysis reveals a pooled global estimate of 64.0% (95% CI: 49.9%-76.0%), with wide variation across nations (range: 18-100%). Notably, the proportion of asymptomatic cases is inversely correlated with the Human Development Index (P=5.6×10−10, R2 = 61.6%), indicating that asymptomatic infections were more prevalent in less developed countries. These results suggest population-level differences in pre-existing immunity or baseline immune responsiveness to SARS-CoV-2.

The second part of the thesis investigates how the physical organization of antigen, specifically, its valency, shapes B cell selection and affinity maturation in germinal centers. I construct a multiscale computational model that captures the interplay of molecular-scale processes (antigen binding, extraction, and signaling) with cellular dynamics (B cell competition, mutation, and selection). Simulations demonstrate that increasing antigen valency enhances the magnitude of the B cell response but often at the expense of its quality. Moreover, there exists a finite valency window that balances signaling strength with discriminatory power, beyond which further increases may dampen immune refinement, highlighting the need for optimization in vaccine design.

Finally, I extend recent theoretical insights into the mechanical regulation of B cell responses by examining how forces during antigen extraction modulate germinal center dynamics. Building on molecular "tug-of-war" models, I discuss how force magnitude influences the balance between affinity discrimination and germinal center stability. These findings provide a physical rationale for the observed valency effects and suggest possible synergies between mechanical cues and antigen multimerization.

Together, this thesis presents new insights into asymptomatic SARS-CoV-2 infections and vaccination strategies. By combining mathematical formalisms, meta-analysis, and multiscale modeling, it advances our understanding of immune variability and informs strategies for disease surveillance and rational vaccine design.