F38. Forecasting the Course of Post-Traumatic Stress Following Emergency Room Admission: A Machine Learning Approach

Abstract

Previous studies identified distinct trajectories of Post-Traumatic Stress Disorder. These studies typically focus on sociodemographic information and severity of potential traumatic brain injury. Moreover, they often investigate these potential predictive factors a long time after the traumatic event took place, thereby limiting the possibility to discern what is a risk factor and what is the consequence of the psychopathology. Thus, there is need to include data comprehensively and to do so immediately after the potentially traumatic event. We examined a broad range of variables in n=338 patients after Emergency Room admission after having experienced a traumatic event. The current work utilizes machine learning to predict post-traumatic stress responses based on biological indicators garnered from electronic medical records including immune markers, heart rate, pulse, along with inter-related endocrine and genetic markers.