Dr. Quindel Jones presents her work to the Biomathematics Seminar.
Title: Mechanistic Models as Hypotheses: Integrating Structure and Data from Sickle Cell Disease to Bacterial Infection
Abstract: Mechanistic models gain scientific value when treated as testable hypotheses rather than fixed descriptions of biology. I illustrate this idea through two modeling studies that differ sharply in how biology and data shape mechanistic structure. I begin with my work on sterile inflammation in sickle cell disease (SCD), where sparse, cross‑sectional clinical data meant the modeling process had to start with biology. I constructed an ODE framework directly from mechanistic understanding of endothelial activation, platelet dynamics, neutrophil behavior, and inflammatory signaling. Because the available clinical data were sparse and cross‑sectional, they served primarily as a validation and parameter‑estimation tool rather than a driver of model structure. This allowed me to identify which parameters most strongly influenced the observed clinical shifts and to characterize the inflammatory pathways most responsible for transitions into vaso‑occlusive crisis. This project reflects a structure‑first approach: biology defines the model, and data provide limited constraints.
In contrast, my current infection work begins with data. Longitudinal murine measurements of bacterial burden, immune activity, and tissue damage allow me to construct and compare multiple mechanistic hypotheses for Klebsiella pneumoniae pneumonia, according to Ganusov’s 2016 framework. By calibrating each model variant and applying AICc‑based model rejection, I use data not simply to validate structure but to actively eliminate unsupported mechanisms through falsification. This strong‑inference approach transforms mechanistic models into testable scientific hypotheses.
Together, these projects illustrate a unified modeling philosophy: mechanistic structure provides interpretability, data provide constraint, and strong inference links the two. The talk will emphasize ODE modeling, nonlinear immune dynamics, and the role of hypothesis‑driven model comparison in understanding complex biological systems.
Date: 2026 February 26, 10:40–11:30
Venue: Little Hall Room 423