Borderline personality disorder classification based on brain network measures during emotion regulation.

Area of Science:

• Neuroscience
• Psychiatry
• Computational Psychiatry

Background:

• Borderline Personality Disorder (BPD) involves emotional dysregulation, often linked to amygdala and prefrontal cortex interactions.
• Empirical evidence on neural mechanisms in BPD remains inconsistent.
• Understanding brain network properties is crucial for BPD research.

Purpose of the Study:

• To systematically evaluate different properties of brain connectivity for detecting BPD.
• To assess the predictive power of task-related functional connectivity in BPD.
• To investigate both task-event dependent (phasic) and task-stable (tonic) connectivity.

Main Methods:

• Functional Magnetic Resonance Imaging (fMRI) during an emotion regulation task.
• Analysis of brain network properties: strength, local efficiency, and participation coefficient.
• Linear support vector machine classification with nested cross-validation.

Main Results:

• Classification accuracy for BPD vs. controls was 55% (not significant).
• Exploratory analysis revealed tonic strength as the highest performing model (62% balanced accuracy).
• The amygdala emerged as a key feature in the tonic strength model.

Conclusions:

• Current fMRI classification methods for BPD may require larger sample sizes.
• Tonic connectivity measures, especially involving the amygdala, warrant further investigation in BPD.
• Future research should focus on network properties of tonic connectivity and potential BPD subgroups.