Degree Type

Dissertation

Date of Award

2013

Degree Name

Doctor of Philosophy

Department

Kinesiology

First Advisor

Gregory J. Welk

Abstract

Accelerometry-based activity monitors have become the standard objective method for assessing physical activity (PA) in field-based research [1]. They are small, non-invasive, easy-to use, and provide an objective indicator of physical activity over extended periods of time. The main advantage from a research perspective is they provide an objective indicator of physical activity behavior, thereby avoiding common sources of error in subjective measurement (e.g., self-report). Because of their storage data capacity, it is possible to monitor behavior over extended periods of time and easy to download the information to a computer for processing. Numerous studies have been published on the reliability and validity of various accelerometry-based physical activity monitors. They have become widely accepted in the field.

Over the years, advances in technology have contributed to dramatic improvements in the sophistication of accelerometry-based monitors. Most monitors today now use 3-dimensional accelerometers with higher sampling rates to provide more detailed information. The use of solid-state construction in most devices has also improved reliability and durability of this class of activity monitor. There have also been many advances in methodologies to process accelerometer data, including more standardization in protocols, better handling of missing data, and the application of complex, pattern recognition techniques to distinguish among various types of movement. These advances have collectively helped advance the science (and practice) of physical activity monitoring, but there are numerous challenges that remain to improve the utility's accuracy of accelerometry-based physical activity monitors.

One of the most challenging problems has been equating output from different accelerometry-based devices. In theory, accelerometry-based devices all measure the same thing (body acceleration). However, there is considerable variability in sensor properties, filtering, and scaling across different monitoring devices. This has made it impossible to directly compare data from competing instruments. While accelerometers internally measure acceleration in g-forces, most commercially-available devices report data using dimensionless units referred to as "counts." Considerable research has been completed to calibrate the various devices against criterion measures, but presently it is not possible to directly equate output from one device to another in a systematic method. In recent years, many new, competing technologies, including built-in accelerometer Smartphones, have been released into the market. These have further compounded the challenge of comparing accelerometry-based devices. In many cases, these devices are released into the marketplace with little or no evidence of their validity.

These accelerometry-based physical activity monitors have been used almost exclusively for research, but advances in technology have led to an explosion of new consumer-based activity monitors designed for use by individuals interested in fitness, health, and weight control. Examples include the BodyMedia FIT, Fitbit, Basis B1, Jawbone Up, NikeFuel band, DirectLife, PAM, and Smartphone applications. The development of these consumer-based monitors and applications has been driven, in large part, by the increased availability of low cost accelerometer technology that came about with the incorporation of accelerometers in the Wii. Refinement of other technologies (e.g., Bluetooth) and increased sophistication of websites and personalized social media applications also spurred the movement. These new accelerometry-based monitors provide consumers with the ability to estimate PA and energy expenditure (EE), and track data over time on a personalized web interface.

Other technologies have also been adapted to capitalize on consumer interest in health and wellness. Pedometers developed originally to measure steps have been calibrated to estimate EE and to store data over time. Geographical positioning system (GPS) monitors, developed primarily for use in navigation, are now marketed to athletes and recreation enthusiasts to monitor speed and EE from activity. Heart rate monitors, originally marketed to athletes, have also been modified and marketed to appeal to more recreational athletes interested in health and weight control. These devices typically provide an easy-to-use web-interface to enable consumers to monitors PA and EE over time.

The increased availability of monitoring technology provides consumers with options for self-monitoring, but these tools may also have applications for applied field-based research or intervention applications designed to promote PA in the population. To date, there is little or no information available to substantiate the validity of these consumer-based activity monitors and accelerometry-based mobile phone applications to assess PA and EE under free-living conditions. It is important to formally evaluate the validity of these various devices so information can be shared with researchers, fitness professionals, and consumers.

The series of papers presented in this dissertation will provide a better understanding of the validity of consumer-based physical activity monitors and also evaluate the potential of estimating energy expenditure using built-in technology in Smartphones. The first study (Chapter 3) specifically evaluated the utility of various consumer-based, physical activity, monitoring tools against indirect calorimetry. A unique aspect of this study is that the consumer-based monitors were also directly compared with results from an Actigraph monitor, the most commonly used research-grade monitor used in the field. The second study (Chapter 4) explored the feasibility and utility of using embedded sensors in smart phones for objective activity monitoring. While consumer-monitors are designed to be convenient and easy to use, it is still somewhat burdensome for individuals to have to wear or carry another device. The embedded sensors in current Smartphones (e.g. accelerometers and gyroscope) may have similar (or better) utility than current research or consumer monitors. However, before this can be done, methods need to be developed to compile and use the raw sensor data. Machine learning techniques are widely used in pattern recognition technology and they have been increasingly used in accelerometry-based monitors to detect underlying patterns in the data. In this second study, machine learning techniques are tested to determine the most optimal way to classify physical activity patterns using Smartphone data. Once this is done it will be possible to develop prediction equations that can convert the raw data into estimates of energy expenditure and/or quantify levels of physical activity (see image below).

The two studies will advance research on physical activity monitoring techniques and specifically determine the feasibility of utilizing embedded sensors in Smartphones to capture physical activity data under free living conditions. A comprehensive literature review is provided in the next section to summarize the progression of research in this area and to explain the methods for the present study.

Copyright Owner

Jung-Min Lee

Language

en

File Format

application/pdf

File Size

147 pages

Share

COinS