Data-Driven Phenotyping of Post-Traumatic Stress Disorder Among Survivors of the 2023 Derna Dam Collapse

Two years after the catastrophic 2023 Derna dam collapse in Libya, the long-term psychological consequences for survivors remain a critical public health concern. Traditional Post-Traumatic Stress Disorder (PTSD) assessments often rely on a single severity metric, which can obscure the heterogeneous nature of trauma responses. This study applies computational modeling to move beyond this monolithic perspective and identify distinct, data-driven trauma phenotypes within the survivor cohort. In this exploratory, cross-sectional analysis of 648 survivors, we employed an unsupervised machine learning framework to uncover latent symptom profiles. The analysis revealed that while 21.5\% of the cohort exhibited high-severity PTSD (Total Score $\geq 34$), the underlying trauma response was not uniform. We identified four robust phenotypes: High Distress (18.8\%), Re-experiencing and Avoidant (26.0\%), Somatic Anxiety and Arousal (30.7\%), and Resilient / Low-Symptom (24.5\%). Phenotype membership was independent of demographic variables, highlighting a latent structure distinct from standard risk factors. These findings demonstrate that the psychological response to the Derna disaster is heterogeneous and can be effectively stratified using computational methods, offering a data-driven foundation for tiered mental health interventions.