OBJECTIVES/GOALS: Fluid boluses are administered to hypotensive, critically ill children but may not reverse hypotension, leading to delay of vasoactive infusion, end-organ damage, and mortality. We hypothesize that a machine learning-based model will predict which children will have sustained response to fluid bolus. METHODS/STUDY POPULATION: We will conduct a single-center retrospective observational cohort study of hypotensive critically ill children who received intravenous isotonic fluid of at least 10 ml/kg within 72 hours of pediatric intensive care unit admission between 2013 and 2023. We will extract physiologic variables from stored bedside monitors data and clinical variables from the EHR. Fluid responsive (FR) will be defined as a MAP increase by 310%. We will construct elastic net, random forest, and a long short-term memory models to predict FR. We will compare complicated course (multiple organ dysfunction on day 7 or death by day 28) between: 1) FRs and non-FRs, 2) predicted FRs and non-FRs, 3), FRs and non-FRs stratified by race/ethnicity, and 4) FRs and non-FRs stratified by sex as a biologic variable. RESULTS/ANTICIPATED RESULTS: We anticipate approximately 800 critically ill children will receive 2,000 intravenous isotonic fluid boluses, with a 60% rate of FR. We anticipate being able to complete all three models. We hypothesize that the model with the best performance will be the long short-term memory model and the easiest to interpret will be the tree-based random forest model. We hypothesize non-FRs will have a higher complicated course than FRs and that predicted non-FRs will have a higher rate of complicated course than FRs. Based on previous adult studies, we hypothesize that there will be a higher rate of complicated course in patients of black race and/or Hispanic ethnicity when compared to non-Hispanic white patients. We also hypothesize that there will be no difference in complicated course when comparing sex as a biologic variable. DISCUSSION/SIGNIFICANCE: We have a critical need for easily-deployed, real-time prediction of fluid response to personalize and improve resuscitation for children in shock. We anticipate the clinical application of such a model will decrease time with hypotension for critically ill children, leading to decreased morbidity and mortality.