Validating Machine -Learned Classifiers of Sedentary Behavior and Physical Activity

Not Applicable

Completed

• Conditions: Sedentary Lifestyle; Physical Activity
• Interventions: Other: Measurement

• Registration Number: NCT01775826
• Lead Sponsor: University of California, San Diego

Brief Summary

The majority of the US population spends most of the day sitting and the we have new scientific evidence that this can contribute to poor health regardless of how much physical activity a person does. However, we do not measure sitting time very accurately and when we ask people to tell us how much they do, their answers are unreliable. Our study will use small sensors to objectively measure when people sit or do physical activity, and we will use sophisticated computational techniques to summarize these movement patterns.

Detailed Description

Not available

Study Timeline

• Study First Submitted: 2013-01-17
• Study First Submitted QC: 2013-01-22
• Study First Posted: 2013-01-25
• Study Start: 2013-03
• Primary Completion: 2016-04
• Study Completion: 2016-04
• Status Verified: 2019-08
• Last Update Submitted: 2019-08-15
• Last Update Posted: 2019-08-20

Recruitment & Eligibility

• Status: COMPLETED
• Sex: All
• Target Recruitment: 225

Inclusion Criteria

Inclusion Criteria for participants 6-17 yr olds:

• provide written parental consent to complete study protocols;
• provide verbal assent to complete study protocols;
• willingness to complete 2 visits to UCSD offices;
• willingness to wear multiple sensor devices on 7 days for 12 hours per day;
• willingness to wear wrist accelerometer on 7 days for 24 hours per day;
• willingness to have their height and weight measured;
• be able to walk unassisted
• able to read and understand study materials in English.

Inclusion Criteria for participants 18-64 yr old:

• provide written consent to complete study protocols;
• willingness to complete 2 visits to UCSD offices;
• willingness to wear multiple sensor devices on 7 days for 12 hours per day;
• willingness to wear wrist accelerometer on 7 days for 24 hours per day;
• complete a survey assessing their demographic characteristics;
• willingness to have their height and weight measured;
• be physically and cognitively able to walk unassisted,
• able to read and understand study materials in English.

Inclusion Criteria for participants 65-85 yr olds:

• provide written consent to complete study protocols;
• correctly answer verbal questions about their comprehension of the informed consent;
• willingness to complete 2 visits to UCSD offices;
• willingness to wear multiple sensor devices on 7 days for 12 hours per day;
• willingness to wear wrist accelerometer on 7 days for 24 hours per day;
• complete a survey assessing their demographic;
• willingness to have their height and weight measured;
• be physically and cognitively able to walk without the assistance of another person (walking aids are permitted)
• able to read and understand study materials in English.

Exclusion Criteria

• unable to ambulate;
• attends a workplace or school on monitoring days that prohibits static images being taken by a SenseCam worn around the neck of the participant;
• pregnancy in second or third trimester.

Study & Design

• Study Type: INTERVENTIONAL
• Study Design: SINGLE_GROUP

Arm && Interventions

Group	Intervention	Description
All Purposes	Measurement	All participants.

Primary Outcome Measures

Name	Time	Method
physical activity behavior classification using study sensors (accelerometers, Sensecam and GPS)	Baseline	Using an annotated data set of SenseCam images in three free-living population subgroups, we will compare sensitivity, specificity and percent agreement between behavioral classifiers derived from: (a) single axis vs. multi axis accelerometers; (b) aggregated movement counts vs. raw acceleration data; (c) hip vs. wrist mounted accelerometers.; Determine (a) the extent to which adding GPS data improves discrimination accuracy over accelerometer only behavior classification (i.e., best classifier resulting from Aim 1); and (b) the extent to which adding GIS data improves discrimination accuracy over accelerometer and GPS behavior classification alone (i.e., best classifier resulting from Aim 2a).

Trial Locations

Locations (1)

UCSD; 🇺🇸 La Jolla, California, United States