Brain Age Prediction in Generalized Anxiety Disorder using a Convolutional Neural Network

Area of Science:

• Neuroimaging
• Psychiatry
• Aging Research

Background:

• Accelerated brain aging is linked to psychiatric disorders.
• Generalized anxiety disorder (GAD) may present markers of accelerated aging.

Purpose of the Study:

• To determine brain predicted age difference (PAD) in individuals with GAD compared to healthy controls (HC).
• To investigate group differences in PAD variability using structural MRI.

Main Methods:

• Utilized a large dataset (3,511 controls, 1,595 GAD, 4,552 HC) from the ENIGMA-Anxiety GAD Working Group.
• Employed a convolutional neural network model with four input modalities for brain age prediction.
• Calculated PAD by subtracting chronological age from predicted brain age.

Main Results:

• Model performance showed consistent accuracy (MAE ≈ 2.95 years) in training and HC testing sets.
• No evidence of accelerated brain aging was found in individuals with GAD.
• Individuals with GAD exhibited significantly greater variability in PAD (Levene's test: W = 442.98, p < .001), particularly those over 25 years old.

Conclusions:

• The study did not find a significant association between GAD and accelerated brain aging (PAD).
• GAD is associated with increased heterogeneity in brain-predicted age, indicating diverse aging trajectories within the GAD group.
• No correlations were observed between PAD and clinical or demographic factors in individuals with GAD.