Predicting Attentional Vulnerability to Sleep Deprivation: A Multivariate Pattern Analysis of DTI Data

Area of Science:

• Neuroimaging
• Cognitive Neuroscience
• Machine Learning

Background:

• Individual differences in sustained attention deficits after sleep deprivation (SD) are significant.
• Previous studies implicated frontal-parietal networks, cortico-thalamic, and inter-hemispheric connections, but lacked individual predictive markers.
• A reliable neuroimaging marker for predicting SD vulnerability at the individual level is currently unavailable.

Purpose of the Study:

• To investigate if white matter (WM) diffusion metrics can predict individual vulnerability to sleep deprivation (SD).
• To apply a machine learning approach to identify neuroimaging markers for SD susceptibility.
• To differentiate between individuals vulnerable and resistant to SD-induced attention deficits.

Main Methods:

• Forty-nine participants underwent the psychomotor vigilance task (PVT) after resting wakefulness and 24-hour SD.
• Participants were categorized as vulnerable or resistant based on PVT lapse changes.
• Diffusion tensor imaging (DTI) data were used to extract WM diffusion metrics (FA, MD, AD, RD).
• A linear support vector machine (LSVM) with leave-one-out cross-validation (LOOCV) was employed to classify participants.

Main Results:

• The LSVM model achieved 83.67% classification accuracy in differentiating SD-vulnerable from SD-resistant individuals.
• Key WM tracts for classification included commissural (superior longitudinal fasciculus), projection (posterior corona radiata, anterior limb of internal capsule), and association pathways (corpus callosum).
• A significant negative correlation was found between changes in PVT lapses and the LSVM decision value.

Conclusions:

• White matter (WM) integrity, particularly in specific fiber tracts, serves as a reliable neuroimaging marker for predicting individual vulnerability to sleep deprivation (SD).
• WM fibers connecting the frontal-parietal attention network, thalamus to prefrontal cortex, and between hemispheres are crucial for predicting SD susceptibility.
• This approach offers a potential tool for identifying individuals at higher risk for attention deficits under sleep deprivation.