Abstract / Summary
Abstract Spinal cord injury (SCI) is a severe neurological condition in which substantial tissue damage develops after the initial trauma through secondary injury mechanisms such as ischemia, inflammation, and metabolic dysfunction. Early management plays a critical role in influencing outcomes, yet current practice lacks data-driven protocols for guiding acute care. This study integrates retrospective clinical analysis, AIassisted data extraction, and computational modelling to support improved SCI management. Retrospective data from 496 patients with traumatic spinal cord injury treated at Landspítali were analysed to investigate demographic patterns, surgical timing, and management factors including mean arterial pressure (MAP). AI-assisted extraction using GPT-4o was evaluated as a proof-of-concept method for identifying predefined clinical variables from Icelandic freetext medical records. In parallel, finite element modelling using anatomically realistic IT’IS Virtual Population models was employed to assess the feasibility of targeted spinal cord cooling. The clinical analysis confirmed known demographic patterns and identified variability in surgical timing and MAP management. AI-assisted extraction successfully identified predefined variables from selected clinical note fragments. Simulation results demonstrated that targeted spinal cooling can achieve clinically relevant reductions in spinal cord temperature while highlighting important physiological and anatomical constraints. Together, these findings provide a data-driven foundation for future SCI protocol development and support continued investigation of neuroprotective strategies in prehospital and early hospital care.