Modeling Health Outcomes Using Accelerometry Data Collected by Different Wearable Devices

• Activity Number: 190 - Novel Approaches for Assessment of Health Outcomes and Multi-Cohort Data Integration Using Wearable Devices in Large-Scale Biomedical Studies
• Type: Topic Contributed
• Date/Time: Monday, August 8, 2022 : 2:00 PM to 3:50 PM
• Sponsor: Section on Statistics in Epidemiology
• Author(s): Ekaterina Smirnova*
• Companies: Virginia Commonwealth University
• Keywords: wearable; accelerometer; meta-analysis; data integration

Abstract

Systematic assessment of population health outcome risks and evaluation of behavioral patterns associated with disease outcomes and mortality is critical for public health and preventive care. A growing number of population level studies (UK Biobank, National Health Examination Survey NHANES) uses accelerometers to associate daily physical activity patterns to health outcomes. The raw accelerometry data can be summarized into a minute-level accelerometry count, monitor independent motion summary (MIMS) or other measure leading to functional data collected over 1440 minutes per each subject day. Systematic analysis of these data from different studies and drawing conclusions about the association between accelerometry and health outcomes presents a number of challenges, including differences in data collection, patient recruitment protocols, accelerometry devices, and minute-level summaries used by each study. We discuss the challenges associated with predictive modeling of the physical activity data collected by different devices and proposed solutions. We illustrate these approaches in the context of mortality outcomes prediction in NHANES 2003-2006 and 2001-2014 studies, which us