OBJECTIVES/GOALS: To investigate digital behavior patterns before cerebrovascular disease (CeVD), we compared accelerometer-measured physical activity (PA) phenotypes of future CeVD patients versus controls in UK Biobank. METHODS/STUDY POPULATION: Accelerometer data from 76,525 eligible participants monitored for 7-days (Jan. 2013-Dec. 2015) was classified into four categories: sedentary, light PA (LPA), moderate-to-vigorous PA (MVPA), and sleep. Covariables and diagnoses were defined using baseline data and patient records. Daily PA patterns associated with incident CeVD were compared to controls using negative binomial regression models. RESULTS/ANTICIPATED RESULTS: Adult participants with future CeVD (n = 2,163) spent 4.4% less time in MVPA (Incident Rate-Ratio (IRR) 0.956; 95% CI = 0.923-0.992; p = 0.016) compared to controls. During 0:00-5:59h (midnight to 5:59AM), future CeVD patients were less likely to sleep (IRR = 0.985; 95% CI = 0.977-0.992; p <0.001) but more likely to be sedentary (IRR = 1.189; 95% CI = 1.098-1.290; p <0.001) or in LPA (IRR = 1.108; 95% CI = 1.015-1.211; p <0.001). In subgroup analyses, decreased MVPA was observed in current/former smokers (IRR = 0.887; 95% CI = 0.819-0.963), males (IRR = 0.931; 95% CI = 0.870-0.997), and the unemployed/retired (IRR = 0.923; 95% CI = 0.856-0.998), an effect more pronounced in depressed patients (p for interaction = 0.044) and prolonged (> 2 hr/day) screen users (p for interaction = 0.018). DISCUSSION/SIGNIFICANCE: The digital phenotype of PA prior to CeVD is characterized by less sleep during 0:00-5:59h and less daily MVPA, demonstrating the utility of accelerometer data in identifying candidates for preventative intervention.