Individualized prediction of dispositional worry using white matter connectivity

Area of Science:

• Neuroscience
• Psychiatry
• Machine Learning

Background:

• Excessive worry is a key symptom in generalized anxiety disorder and other psychiatric conditions.
• Individualized prediction of worry propensity is crucial for clinical assessment of symptom severity.

Purpose of the Study:

• To predict dispositional worry using machine learning based on white matter (WM) tract microstructural integrity.
• To identify potential neuromarkers for clinical worry assessment.

Main Methods:

• A multivariate machine learning approach was employed.
• Dispositional worry was predicted from the microstructural integrity of white matter (WM) tracts.

Main Results:

• The machine learning model successfully decoded individual dispositional worry scores from WM microstructure.
• Key WM tracts involved included the posterior limb of the internal capsule, anterior corona radiate, cerebral peduncle, and corticolimbic pathways.
• Model performance metrics: mean absolute error = 10.46, root mean squared error = 12.82, prediction R2 = 0.17 (all p < 0.001).

Conclusions:

• This study elucidates potential neuromarkers for clinical assessment of worry symptoms across diverse psychiatric disorders.
• Identified WM pathways enhance the understanding of the neurobiological mechanisms underlying worry propensity.