Detecting and intervening in rumination before the onset of clinical episodes is crucial, highlighting the need for objective markers and predictive models. Resting-state functional magnetic resonance imaging (rsfMRI) has shown promise in developing brain-based markers for rumination, particularly within the brain’s default mode network (DMN).
The DMN has been consistently implicated in rumination and is also associated with other mental health disorders characterized by dysfunctional thoughts and emotions.
Heterogeneity within the DMN and its Subsystems
The DMN is a complex network, both in terms of its anatomy and function. It comprises interacting subsystems that support the cognitive processes involved in the construction of thoughts and emotions. One such subsystem, the medial temporal system, facilitates low-level construal of mental experiences, imbuing mental simulations with rich spatial, temporal, and perceptual details.
In contrast, the dorsal medial system supports high-level reflective processes, allowing individuals to consider the broader implications and significance of their thoughts, emotions, and external stimuli. The dorsal medial prefrontal cortex (dmPFC), a core region within the DMN, is proposed to play a crucial role in ruminative thinking, as it modulates connectivity related to depression and exhibits heightened connectivity in major depressive disorder.
The Importance of Dynamic Functional Connectivity
The ruminative style is characterized by persistent and difficult-to-release negative self-focused thoughts that persist over time. Understanding this temporal aspect of rumination is crucial, and it is hypothesized that the variance of dynamic functional connectivity within the DMN serves as a predictor of rumination.
Few studies have explored the relevance of dynamic connectivity variance to rumination, and no previous study has developed dynamic connectivity-based predictive models to predict rumination in new individuals.
Goals of the Study
The present study aims to apply dynamic connectivity-based predictive modeling to multiple independent datasets, including subclinical and clinical samples, to answer several questions. Firstly, can a generalizable predictive model of rumination be developed using rsfMRI-based dynamic functional connectivity within subclinical samples? Secondly, which functional connections contribute significantly to the prediction of rumination?
Finally, can the model predict depression scores in a clinical sample? To address these questions, the study utilizes three independent rsfMRI datasets from subclinical samples for training, validation, and independent testing of the predictive models. Multiple models are developed to predict the different subscales of the Ruminative Response Scale (RRS), including brooding, depressive rumination, and reflective pondering.
Methodology and Results
The models are tested on validation and independent test datasets to identify those that generalize across multiple datasets. Additionally, the virtual lesion method is employed to identify important features from the original full model. Finally, the refined model, comprising the identified key features, is tested on a separate clinical dataset consisting of individuals with major depressive disorder to predict their depression scores measured with the Beck Depression Inventory-II.
Conclusion
The study successfully develops a dynamic connectivity-based predictive model of rumination, highlighting the importance of the dmPFC and functionally connected regions in the psychological processes underlying trait rumination. This model holds potential for evaluating rumination in both subclinical and clinical populations, providing a valuable tool for identifying individuals at risk for mental illness and guiding targeted interventions.
reference link :https://www.nature.com/articles/s41467-023-39142-9
