Temporal Bayesian Network modeling approach to evaluating the emotional cascade model of borderline personality disorder

Abstract

Theoretical models of personality disorders can be complex and multifaceted, making it difficult to validate such models in a comprehensive, empirical fashion. One such model of borderline personality disorder (BPD) is the emotional cascade model (Selby & Joiner, 2009), which has garnered empirical support in piecemeal fashion but has not been examined in a gestalt fashion. One way to test comprehensive models of personality pathology is with Temporal Bayesian Network (TBN) modeling, in which the relations between multiple subcomponents of a model can be specified and examined over a dynamic time frame, allowing for the modeling of positive feedback processes in addition to comprehensive model utility. In this study, we applied TBN modeling to examine the emotional cascade model in a sample of adolescents and young adults who actively self-injure, including those with BPD. TBN modeling was applied to ecological momentary assessment data provided via participant smartphone assessments for a period of 2 weeks. TBN analysis suggested that the emotional cascade model has considerable predictive utility, demonstrating substantial accuracy in predicting BPD diagnosis (with accuracy estimates around 90%) and momentary prediction of rumination, negative emotion, and dysregulated behaviors (with accuracy estimates consistently above 70% and reaching up to 100%, depending on the level of momentary prediction specificity). These findings provide support and validity to the notion that BPD may emerge from a dynamic interplay between emotional cascades and dysregulated behaviors. Implications of TBN modeling of BPD and personality disorders, in general, are discussed.