CMStatistics 2020: Start Registration
View Submission - CMStatistics
B0965
Title: Estimation of free-living walking cadence from wrist-worn sensor accelerometry data Authors:  Marta Karas - Johns Hopkins Bloomberg School of Public Health (United States) [presenting]
Jacek Urbanek - Johns Hopkins University (United States)
Ciprian Crainiceanu - Johns Hopkins University (United States)
Jonas Dorn - Novartis (Switzerland)
Abstract: Walking and gait parameters have become increasingly important in epidemiological and clinical studies. Indeed, 3 out of 7 submissions to the FDA for eCOA qualifications of digital endpoints are quantifying gait parameters. Recent evidence suggests that observations collected in a free-living environment are complementary to traditional, lab- and clinic-based walking measurements. Sub-second-level actigraphy data can provide a detailed description of human movement. Despite the growing need, the number of publicly available methods to derive gait from high-density accelerometry data collected by wrist-worn devices is limited. We propose an extension of ADEPT, a pattern-matching method, to segment individual walking strides in sub-second-level accelerometry data collected in a free-living environment using a wrist-worn sensor. We evaluate the method on 4-week observation data from 30 people with and 15 people without arthritis. We show that daily walking cadence is significantly associated with general mental health, social functioning, and role physical scores reported via SF-36. We provide open-source software and out-of-study sample data examples online.