Understanding Transient Network Effects Triggered by Spontaneous Events during Sleep with Biophysical and Statistical Models

Abstract

Sleep and its major functions require the precise coordination of transient mechanisms at multiple spatiotemporal scales, which are reflected in neural signals by the spontaneous occurrence of transient neural events. At the systems level, the network dynamics determine how and when such events are generated. Conversely, the occurrence of events influences the underlying network properties. We are interested in the interplay between the events and the underlying network mechanisms to address the potential functions of transient activity during sleep.

In this thesis, we achieve this goal with both biophysical and statistical modelling approaches. In the first project, we build a biophysically interpretable model of PGO waves based on network mechanisms at different levels and make prediction about PGO-triggered plasticity. In the second project, we apply statistical models that take manageable forms and estimate peri-event dynamics from data. Although this does not explicitly account for detailed biological mechanisms, we show that they can provide qualitative insights through causality analysis. Finally, by combining models of both perspectives, we propose a hybrid framework that integrates experimental data with models to estimate biologically meaningful parameters and recover hidden brain states.