Classifying Vulnerability to Sleep Deprivation Using Resting-State Functional MRI Graph Theory Metrics

Area of Science:

• Neuroscience
• Cognitive Science
• Medical Imaging

Background:

• Sleep deprivation (SD) is prevalent, causing cognitive deficits with significant individual variability.
• Understanding neural predictors of SD vulnerability is crucial for targeted interventions.
• Previous research has identified trait-like cognitive differences but limited neuroimaging-based predictive models.

Purpose of the Study:

• To investigate the potential of resting-state functional magnetic resonance imaging (fMRI) and graph theory to predict vulnerability to sleep deprivation.
• To classify individuals as vulnerable or resilient to SD based on neuroimaging data.
• To identify specific brain networks and regions associated with SD vulnerability.

Main Methods:

• Collected resting-state fMRI and psychomotor vigilance task (PVT) data from 60 healthy participants under normal and sleep-deprived conditions.
• Utilized graph-theory-based degree centrality (DC) to analyze brain network organization.
• Employed support vector machine (SVM) with leave-one-out cross-validation for classification of SD vulnerability.

Main Results:

• Significant widespread changes in DC were observed after SD, indicating brain network reorganization.
• SVM analysis achieved an 84.75% classification accuracy for differentiating vulnerable from resilient individuals.
• Key brain areas contributing to classification included the sensorimotor network, default mode network, and thalamus.

Conclusions:

• Resting-state network measures, particularly DC, show promise in predicting vulnerability to sleep deprivation.
• Machine learning algorithms combined with neuroimaging data can effectively screen individuals susceptible to SD.
• Thalamic DC changes correlated with PVT performance deficits after SD, highlighting its role.