A Non-linear Predictive Model of Borderline Personality Disorder Based on Multilayer Perceptron

Area of Science:

• Psychiatry
• Psychology
• Mental Health

Background:

• Borderline Personality Disorder (BPD) is a complex mental illness often linked to trauma and dissociative symptoms.
• Individual differences in temperament and character may predispose individuals to BPD.
• Understanding the interplay of these factors is crucial for diagnosis and treatment.

Purpose of the Study:

• To investigate the predictive value of dissociative symptoms and personality traits on BPD diagnosis.
• To explore the relationship between trauma, character fragility, and BPD development.
• To utilize advanced statistical modeling to analyze these complex interactions.

Main Methods:

• A sample of 602 outpatients receiving mental health services was assessed.
• BPD diagnosis was determined using a semi-structured interview for DSM-IV-TR personality disorders.
• Dissociative symptoms and personality traits were measured using the Dissociative Experiences Scale and Cloninger's Temperament and Character Inventory, respectively.
• A multilayer perceptron neural network model was employed to analyze the data.

Main Results:

• The statistical model achieved an 89% accuracy in predicting BPD.
• Dissociative symptoms demonstrated a greater influence on BPD diagnosis than character traits.
• Harm Avoidance and Reward Dependence (temperament) and Cooperativeness and Self-Directedness (character) showed significant associations with BPD.

Conclusions:

• Both psychological factors (temperament, character) and environmental factors (trauma) contribute to BPD development.
• Dissociative symptoms play a significant role in the manifestation of BPD.
• These factors interact non-linearly, leading to the maladaptive behaviors characteristic of BPD.