Abstract

PURPOSE: Test the accuracy of pattern-recognition based physical activity (PA) modeling algorithms in overweight Hispanic youth.
METHODS: Youth visited our PA laboratory for 4 sessions (different days and times) and completed 9 structured activities per session (7 minutes per activity), wearing 2 tri-axial accelerometers and one heart rate monitor. Activities, based on an enhanced SOFIT protocol, were: lying down, sitting, sitting fidgeting, standing, standing fidgeting, standing playing an active computer game (Wii), slow walking, brisk walking, and running. Data from 3 sessions was used to train an activity recognition model. Data from session 4 was used to test the model. A fusion system to recognize these activities that we previously developed in a mixed adult-youth sample was tested in a sample of overweight Hispanic youth. The system classifies multimodal temporal and cepstral features using a …

Date

2010

Authors

Ming Li, Adar Emken, Shri Narayanan, Gautham Thatta, Harshvardhan Vathsangam, Gaurav Sukhatme, Urbashi Mitra, Murali Annavaram, Spruijt-Metz Donna

Journal

Medicine & Science in Sports & Exercise

Volume

42

Issue

5

Pages

485

Publisher

LWW