
Head and neck trauma are common among intimate partner violence (IPV) survivors and contribute substantially to traumatic brain injury (TBI). Injury surveillance systems often fail to capture the behavioral mechanisms through which IPV-related injuries occur, limiting etiological classification. Violence-related injury records from the National Electronic Injury Surveillance System (NEISS) were analyzed. Regular expression text classification identified IPV-related cases referencing intimate relationships, and analyses were restricted to head and neck injuries. A large language model (LLM) extracted standardized information describing injury action, cause, mechanism, object, and location. Latent Dirichlet Allocation (LDA) topic modeling identified recurring injury patterns within each domain. Distinct pathways to IPV-related head and neck injury were identified. The most prevalent typologies included accidental falls or slips during interpersonal conflict (38.6%), partner-inflicted physical assault (25.1%), and intentional self-inflicted injury during arguments (19.9%). Additional pathways included blunt object impacts (9.9%), environmental or chemical exposures (2.3%), and vehicle-related injuries (1.7%). The application of AI-assisted analytic tools to ED narratives improves identification of IPV-related head and neck injury mechanisms not captured by NEISS structured surveillance fields.
Session: Injury Epidemiology, Trauma Systems, and Health Care Utilization Posters
Program: Injury Control and Emergency Health Services
Date: Monday, November 2, 2026 | 10:30 AM – 11:30 AM