Master's Thesis Defense: "Effects of functional network model definition on biomarker outcome prediction"
Abstract: Adult resting state networks (RSN) have been reliably indexed by various analysis methods. However, using adult RSN in pediatric functional connectivity (FC) datasets may result in poor model fit which might limit the generalizability of the results. Pediatric research studies have chosen either to report FC results using either adult networks or pediatric networks. No studies to date have quantified the difference in prediction accuracy and reliability of pediatric FC when using adult vs pediatric network models. In this talk, we demonstrate that age-specific network models are crucial for producing biologically interpretable, accurate, reliable, and reproducible predictions. We will further explain how we use machine learning models to quantify these results. This work has important implications for the prediction of clinical outcomes in developing populations and highlights the need for standardized system-level atlases in pediatric populations.
Hosts: Soumendra Lahiri and Muriah D. Wheelock