Physical activity intensity can be accurately monitored ...

European Journal of Sport Science

ISSN: 1746-1391 (Print) 1536-7290 (Online) Journal homepage: http://www.tandfonline.com/loi/tejs20

Physical activity intensity can be accurately monitored by smartphone global positioning system ‘app’ Brett Ashley Gordon, Lyndell Bruce & Amanda Clare Benson To cite this article: Brett Ashley Gordon, Lyndell Bruce & Amanda Clare Benson (2015): Physical activity intensity can be accurately monitored by smartphone global positioning system ‘app’, European Journal of Sport Science, DOI: 10.1080/17461391.2015.1105299 To link to this article: http://dx.doi.org/10.1080/17461391.2015.1105299

Published online: 27 Oct 2015.

Submit your article to this journal

Article views: 21

View related articles

View Crossmark data

Full Terms & Conditions of access and use can be found at http://www.tandfonline.com/action/journalInformation?journalCode=tejs20 Download by: [La Trobe University]

Date: 01 November 2015, At: 15:27

European Journal of Sport Science, 2015 http://dx.doi.org/10.1080/17461391.2015.1105299

ORIGINAL ARTICLE

Physical activity intensity can be accurately monitored by smartphone global positioning system ‘app’ BRETT ASHLEY GORDON1,2,∗

, LYNDELL BRUCE2, & AMANDA CLARE BENSON2

Discipline of Exercise Physiology, La Trobe Rural Health School, La Trobe University, Bendigo, VIC, Australia & 2Discipline of Exercise Sciences, School of Medical Sciences, RMIT University, Melbourne, VIC, Australia

Downloaded by [La Trobe University] at 15:27 01 November 2015

1

Abstract Monitoring physical activity is important to better individualise health and fitness benefits. This study assessed the concurrent validity of a smartphone global positioning system (GPS) ‘app’ and a sport-specific GPS device with a similar sampling rate, to measure physical activity components of speed and distance, compared to a higher sampling sport-specific GPS device. Thirty-eight (21 female, 17 male) participants, mean age of 24.68, s = 6.46 years, completed two 2.400 km trials around an all-weather athletics track wearing GPSports Pro™ (PRO), GPSports WiSpi™ (WISPI) and an iPhone™ with a Motion X GPS™ ‘app’ (MOTIONX). Statistical agreement, assessed using t-tests and Bland–Altman plots, indicated an (mean; 95% LOA) underestimation of 2% for average speed (0.126 km·h−1; –0.389 to 0.642; p < .001), 1.7% for maximal speed (0.442 km·h−1; –2.676 to 3.561; p = .018) and 1.9% for distance (0.045 km; –0.140 to 0.232; p < .001) by MOTIONX compared to that measured by PRO. In contrast, compared to PRO, WISPI overestimated average speed (0.232 km·h−1; –0.376 to 0.088; p < .001) and distance (0.083 km; –0.129 to –0.038; p < .001) by 3.5% whilst underestimating maximal speed by 2.5% (0.474 km·h−1; –1.152 to 2.099; p < .001). Despite the statistically significant difference, the MOTIONX measures intensity of physical activity, with a similar error as WISPI, to an acceptable level for population-based monitoring in unimpeded open-air environments. This presents a low-cost, minimal burden opportunity to remotely monitor physical activity participation to improve the prescription of exercise as medicine. Keywords: GPS, aerobic exercise, validity, smartphone application, activity monitor

Introduction The importance of physical activity on health and disease prevention is unequivocal (Paffenbarger, Hyde, Wing, & Hsieh, 1986); however, few adults (approximately 10–40%) participate in sufficient amounts of physical activity (Australian Bureau of Statistics, 2013; Tucker, Welk, & Beyler, 2011), particularly if they are obese (Davis, Hodges, & Gillham, 2006). The ability to monitor individual physical activity behaviour is positively associated with the volume of physical activity completed (Son, Kerstetter, Mowen, & Payne, 2009; Umstattd & Hallam, 2007). It is therefore important to assess and monitor community-based physical activity to determine whether individuals are meeting and adhering to current physical activity and exercise recommendations to obtain the associated health benefits (Haskell et al., 2007; Nelson et al., 2007; Wilmot ∗

et al., 2012; Zinman, Ruderman, Campaigne, Devlin, & Schneider, 2004), and enable appropriate recommendations on how increased activity could be attained and adhered to (Bandura, 2004; Dishman, Vandenberg, Motl, Wilson, & DeJoy, 2010). Physical activity participation can be monitored with Global Positioning System (GPS), which are more accurate than self-report (Badland, Duncan, Oliver, Duncan, & Mavoa, 2010; Duncan & Mummery, 2007; Stopher, Fitzgerald, & Xu, 2007), and less time-consuming than direct observation (Maddison & Ni Mhurchu, 2009). Additionally, GPS provides important contextual information on physical activity completed outdoors or when satellite signals can be obtained through glass windows, which pedometers and accelerometers cannot (Krenn,

Correspondence: B. A. Gordon, Discipline of Exercise Physiology, La Trobe Rural Health School, La Trobe University, PO Box 199, Bendigo, VIC 3552, Australia. E-mail: [email protected]

© 2015 European College of Sport Science


2

B. A. Gordon et al.

Titze, Oja, Jones, & Ogilvie, 2011). This contextual information provides the location and terrain of where the activity was performed, that when combined with the spatial data collected can provide specific insights to the activity session and profile for the individual. Spatial data collected from GPS, including speed and distance, are more specific than spatial data collected from accelerometers, which only give the number and frequency of physical activity counts (Pedišić & Bauman, 2015). However, the ability to fix to satellites is required for accurate GPS data to be recorded. Static accuracy of many GPS devices is good (within 5 m of actual location) in open areas (Duncan et al., 2013; Rodriguez, Brown, & Troped, 2005), but substantially worse in high-rise areas (approximately 50 m from actual position) (Duncan et al., 2013). Similarly, dynamic accuracy is increased when greater positional logging occurs in open areas (average less than 5 m from actual location) compared to urban canyons (average of 9.3 m from actual location) (Schipperijn et al., 2014). While tracking of distance can be affected by built-up areas restricting the ability to fix with satellites, other exercise variables such as speed (intensity) have not been systematically evaluated outside sporting arenas (Coutts & Duffield, 2010; Duffield, Reid, Baker, & Spratford, 2010; Johnston et al., 2012). The use of GPS to monitor physical activity in health research is becoming increasingly prevalent. Commercially available GPS units are popular amongst sporting teams and accurately measure distance and speed (Brewer, Dawson, Heasman, Stewart, & Cormack, 2010; Coutts & Duffield, 2010; Gray, Jenkins, Andrews, Taaffe, & Glover, 2010; Jennings, Cormack, Coutts, Boyd, & Aughey, 2010), although due to cost, they are unlikely to be feasible for large-scale measurement and monitoring of population-based research or community-based exercise interventions (Maddison & Ni Mhurchu, 2009). When GPS has been utilised within community-based research, it has investigated active transport and the environmental role in promoting physical activity (Krenn et al., 2011; Maddison & Ni Mhurchu, 2009); not its ability to measure variables associated with community-based physical activity and exercise. Smartphone ownership is substantial in westernised countries (Pew Internet Project, 2015; Sensis eBusiness Report, 2014). Smartphones now have GPS included as a standard feature; however, an appropriate application (‘app’) is required to utilise the data. The cost of GPS ‘apps’ ranges from free to tens of pounds and can accurately and reliably record exercise distance and average speed (Benson, Bruce, & Gordon, 2015); however, their ability to

determine exercise intensity compared with that measured by sport-specific GPS devices is unclear. Therefore, the aim of this study was to assess the concurrent validity of a common iPhoneTM ‘app’ and sports-specific GPS device (GPSports WiSpi™), with similar sampling frequencies, for the purpose of monitoring physical activity speed and distance compared to a higher sampling commercially available sport-specific GPS device (GPSports Pro™). Methods Study participants This study was approved by the University’s Human Research Ethics Committee and conformed to the principles of the Declaration of Helsinki. Forty apparently healthy male and female individuals, aged 18–55 years, consented to participate in this study. Participants completed a health-risk screening questionnaire and were excluded if any musculoskeletal injuries or medical conditions contraindicated for walking or running were identified. The mean age of included participants was 24.68, s = 6.46 years old, and they were 170.96, s = 6.99 cm tall and had a body mass of 69.45, s = 8.81 kg. Height and body mass were measured using standardised procedures (Gore, 2000) before completing the experimental protocol.

Experimental protocol A cross-sectional study design was used to assess the concurrent validity of a low-sampling smartphone GPS application and a sport-specific GPS unit with a higher sampling sport-specific GPS unit. Participant’s wore the GPSports ProTM GPS unit (PRO) on the middle of their back in a standard bib, a GPSports WiSpiTM GPS unit (WISPI) in an armband on their upper right arm and the iPhoneTM using the Motion X GPSTM ‘app’ (MOTIONX) in an armband on their upper left arm while completing six laps of a 400 m outdoor all-weather running track. The MOTIONX was chosen on the basis that at the time of the study, it was available as a free download, which would increase the likelihood of an individual using the ‘app’ as opposed to one they had to pay to download, and provided a function to export the data from the physical activity session. The MOTIONX (Motion X GPSTM, Fullpower Technologies, Inc, Santa Cruz, CA, USA) used the inbuilt iPhoneTM location services API to obtain its sampling rate (which is likely to be ≤1 Hz), while the WISPI and PRO sampled at 1 and 5 Hz, respectively (GPSports, ACT, Australia). A higher sampling rate has been reported to enable data to be recorded


Smartphone GPS ‘app’ physical activity measurement more accurately when frequent changes of direction are performed, but should not be a factor when monitoring motion completed in a straight line (Coutts & Duffield, 2010; Duffield et al., 2010). Each participant completed two trials of the six lap protocol (2.400 km), consisting of two laps walking, three laps of walking and jogging (100 m walk, 100 m light jog, 100 m walk, 100 m fast jog) followed by a final lap of walking, to replicate the transition between intensities that typically occur during aerobic-type physical activity completed by non-athletic populations for health or fitness purposes. All units were turned on at the start and off at the end of each trial at the same location on the 400 m track. GPS units require time for initialisation prior to beginning movement to ensure that the units obtain a satellite fix. Therefore, the zero speed point for each unit was found and confirmed with manually recorded start and end times for each trial with data removed prior to and after the end of each trial. The technology was deemed to have failed on two occasions by two independent reviewers (one occasion with the WISPI during trial one and once with the MOTIONX during trial two). These two individuals therefore had all of their data excluded from analysis, meaning data from 38 (21 female, 17 male) participants were analysed across 76 trials. The PRO and WISPI units were downloaded in a standard docking station and data analysed using the GPSportsTM software (GPSports Team AMS Release R1 2011.8, Canberra, ACT, Australia) at

3

the completion of the trials. Ending a trial using the MOTIONX created a ‘track’ that was emailed to the researchers for later analysis.

Statistical analysis Data from both trials were utilised to assess the validity of the devices from the 76 trials. All data were analysed using IBM SPSS Statistics 21 for Windows (SPSS, Chicago, IL) with the exception of Bland–Altman analysis performed with GraphPad Prism 5.01 for Windows. Data are presented as mean and SD unless otherwise specified, with significance set at p ≤ .05. Concurrent validity. The absolute validity of these sport-specific devices has been established (Benson et al., 2015; Coutts & Duffield, 2010; Gray et al., 2010; Johnston et al., 2012), which demonstrates that the PRO is a suitable criterion to compare the MOTIONX- and WISPI-measured average speed, maximal speed and distance. Validity was examined via paired t-tests. To assess for bias and statistical agreement between the PRO and MOTIONX and the PRO and WISPI, Bland–Altman plots were analysed. Validity was further assessed with Pearson correlation for self-selected average and maximal speed, and distance. The magnitude of the difference was calculated in three ways: (i) the validity coefficient of variation (CV) was calculated by dividing the

Table I. Concurrent validity of average speed, maximal speed and distance for each GPS unit compared to GPSports Pro™

Average speed (km·h −1) mean (SD) Bias ± 95% LOA SD of difference Validity CV (%) P Effect size (Cohen’s d ) Maximal speed (km·h −1) mean (SD) Bias ± 95% LOA SD of difference Validity CV (%) p Effect size (Cohen’s d ) Distance (km) mean (SD) Bias ± 95% LOA SD of difference Validity CV (%) p Effect size (Cohen’s d )

PRO

WISPI

MOTIONX

6.724 (0.655) – – – –

6.955 (0.663) –0.232 ± 0.144 0.011 0.507