Personality biomarkers of pathological gambling: A machine learning study

Area of Science:

• Psychiatry and Behavioral Science
• Computational Psychology
• Machine Learning in Healthcare

Background:

• Maladaptive personality traits are recognized risk factors for Gambling Disorder (GD).
• Identifying specific personality profiles associated with GD is crucial for understanding its etiology.

Purpose of the Study:

• To apply artificial intelligence (AI) for extracting personality predictors of GD.
• To develop a machine learning model for individual-level GD diagnosis using personality data.

Main Methods:

• Utilized Classification and Regression Trees (CART) algorithm.
• Employed the Five-Factor Model of Personality (NEO-PI-R) data from 40 GD patients and 160 healthy controls.
• Built a classification model to differentiate GD patients from controls.

Main Results:

• The classification model achieved an Area Under the Curve (AUC) of 77.3% (p<0.0001).
• The algorithm identified predictive patterns based on openness, neuroticism, and conscientiousness sub-facets.
• Demonstrated successful discrimination between individuals with and without GD at an individual level.

Conclusions:

• This study is the first to integrate behavioral data with machine learning for GD personality feature extraction.
• The findings provide a proof-of-concept for AI-driven GD diagnosis.
• The identified multivariate personality profiles may aid in assessing vulnerability to GD in clinical settings.