Change point based dynamic functional connectivity estimation outperforms sliding window and static estimation for classification of early mild cognitive impairment in resting-state fMRI

The preprint operates within the dominant paradigm of using functional magnetic resonance imaging (fMRI) and network neuroscience to study brain function and disorders. Specifically, it focuses on the sub-paradigm of dynamic functional connectivity (DFC) analysis, which has gained traction in recent years as a way to capture the time-varying nature of brain networks.

This preprint falls under the "Normal Science" stage of Kuhn's cycle. It works within the established paradigm of fMRI-based network neuroscience and DFC analysis. The research seeks to refine existing methods within this paradigm by comparing the performance of different DFC estimation techniques (sliding window and change point based) for classifying early mild cognitive impairment (eMCI). The preprint doesn't challenge the underlying assumptions of the paradigm, but instead aims to improve the accuracy and interpretability of results within it. While the authors highlight the limitations of the widely used sliding window approach and advocate for change point based methods, this represents a refinement of existing tools, not a paradigm shift. The core concepts of the field, such as using fMRI to study brain networks and the importance of DFC, remain unchallenged. There's a potential hint of "Model Drift", as the paper subtly suggests limitations of the sliding window DFC approach and explores the change-point method as a better fit for individual brain dynamics. This, however, doesn’t constitute a full "Model Drift", as it doesn't demonstrate widespread failure of the existing model, nor does it propose a radically new alternative model. It explores a less common method that still aligns with the overarching paradigm. Thus, "Normal Science" with a possible lean towards, but not fully reaching, "Model Drift" is the most appropriate classification.