Mapping Post-Traumatic Stress Disorder: A Proof of Concept for Network-Based Case Formulation

Background and Objective. The network approach to psychopathology views mental disorders as dynamic interactions between symptoms, challenging traditional latent-cause models. We applied this approach to individuals with PTSD, using ecological momentary assessment (EMA) data to build personalized contemporaneous and directed symptom networks. Our primary aim was to assess whether these network-derived formulations aligned with clinician-generated cognitive formulations. We also tested whether cognitive variables predicted affective variables over time, consistent with the Ehlers and Clark (2000) cognitive model of PTSD. Method. Five participants completed 14-21 days of EMA (response rate=90.2+5.48%), yielding 348 observations total. Items were tailored to each patient’s lived experience and mapped to Ehlers and Clark components. Written formulations from patients and clinicians were compared to EMA-based networks generated by three network algorithms.Results. Network-derived formulations shared some features with patient and clinician-derived formulations but often diverged in complexity and specific predictions. Data-driven networks had good precision (0.64-0.86) but low sensitivity (0.31-0.35), showing better agreement with patient self-formulations than with clinician models. Evidence for lead-lag effects between cognitive and affective variables was limited. Trauma memory emerged as the most central component, with its association to sense of current threat and matching triggers being the most prominent. Conclusion. Network analysis may refine clinical hypotheses but has limitations in replicating comprehensive formulations. Instead, we recommend its use for clinical hypothesis testing of specific predictions. Further studies, including randomized trials, are needed to test its clinical utility and contribution to treatment outcomes.