Toward a hybrid assessment framework for adolescent borderline personality disorder: a mini review of personality functioning, digital biomarkers, and AI-supported assessment.

Borderline personality disorder (BPD) in adolescence is increasingly recognized as a valid and clinically meaningful diagnosis, yet it remains frequently underdetected and undertreated. Diagnostic uncertainty often arises from overlap between typical developmental features, such as emotional reactivity, identity exploration, and fluctuating peer relationships, and the pathological instability characteristic of BPD. Stigma and clinician hesitancy to diagnose before age 18 further delay intervention, despite evidence that early identification improves outcomes. Dimensional diagnostic frameworks, including the DSM-5 Alternative Model for Personality Disorders and ICD-11, define personality pathology through impairments in levels of personality functioning (LPF) across self and interpersonal domains. Validated adolescent instruments such as the LoPF-Q 12-18 and AIDA operationalize these constructs, but rely on static and retrospective self-report, limiting their ability to capture rapid, context-dependent shifts in adolescent BPD. This mini review synthesizes diagnostic challenges, evaluates LPF-based assessment tools, and highlights the need for approaches that better capture dynamic variability. Digital phenotyping and artificial intelligence (AI) offers promising adjuncts to traditional assessments. Candidate digital biomarkers from smartphones and wearables, including speech patterns, activity data, and physiological signals, may provide continuous and ecologically valid indicators relevant to personality functioning. Rather than presenting a validated diagnostic framework, this mini review proposes a conceptual direction for future hybrid assessment integrating LPF-based tools, structured clinical interviews, ecological momentary assessment, and candidate digital biomarkers.