Detecting Narcissistic Personality Disorder (NPD): A Hybrid Regex and NLP based AI Approach with Phase-Aware Classification

Narcissistic Personality Disorder (NPD) is considered one of the three malevolent personality traits comprising the ’Dark Triad’ alongside Machiavellianism and Psychopathy. Recent advances in computational psycholinguistics have demonstrated the potential of natural language processing (NLP) for the detection of personality disorders. To address the complexities of detecting nuanced linguistic patterns associated with NPD and abuse cycles, hybrid models that integrate rule-based and deep learning approaches have been proposed. Our approach synergises a transparent Regex based system for explicit markers with a fine-tuned, domain-adapted BERT model for implicit, contextual patterns. Crucially, we validated this hybrid system through a rigorous three-stage process, demonstrating a replicable methodology that successfully bridges the domain gap to real-world proxy data for toxic online discourse. This work provides a robust foundation for developing computational tools to aid researchers and clinicians in analysing textual data for patterns relevant to narcissistic dynamics.