Morten Bilde Simonsen1, Mikkel Jacobi Thomsen2, Rogerio Pessoto Hirata2. 1. Sport sciences, Department of Health Science and Technology, Aalborg University, Fredrik Bajers Vej 7, DK-9220 Aalborg East, Denmark; Centre for Clinical Research, North Denmark Regional Hospital, Bispensgade 37, DK-9800 Hjoerring, Denmark. Electronic address: mbsi@hst.aau.dk. 2. Sport sciences, Department of Health Science and Technology, Aalborg University, Fredrik Bajers Vej 7, DK-9220 Aalborg East, Denmark.
Abstract
BACKGROUND: Commercially available physical activity trackers are very popular in the general population and are increasingly common in clinical and research settings. The marketfor activity trackers are rapidly expanding, requiring them to be validated on an ongoing basis. Different approaches have been used for validating these devices. Studies using treadmills shows good step-counting accuracy although test performed in field tests settings are limited. RESEARCH QUESTION: Does step-counting validity differ between a field test and a treadmill protocol for different types of activity trackers? METHODS: Thirty healthy subjects participated in this study, mean age was 28.2 (± 4.33) years, body mass 78.9 (± 12.9) kg, and height 178.5 (± 9.7) cm. A treadmill protocol with three different walking speeds (2, 3 and 4 km/h) and a 982 m field test was used. During the tests, participants' feet were filmed using a waist-mounted camera. The number of steps were extracted from the video data and used for comparison with four different step counters: a) Polar M200; b) Polar A300; c) Dunlop pedometer; d) Samsung Galaxy S9 smartphone. Validity and agreement determined was determined with the use of Bland-Altman plot and Spearman's correlation. RESULTS: Validity was higher for the field test compared to the 4 km/h treadmill test for all tested devices. The smartphone was the most accurate in terms of error, validity and agreement for both the treadmill and field test. All devices performed poorly for the 2 km/h treadmill test and only the smartphone performed well at 3 km/h. SIGNIFICANCE: The results of this study show that step counting validity and error obtained during treadmill walking is not similar to a field test. Future validation studies of activity trackers should consider this when designing a protocol. The smartphone had the lowest mean bias during the field test.
BACKGROUND: Commercially available physical activity trackers are very popular in the general population and are increasingly common in clinical and research settings. The marketfor activity trackers are rapidly expanding, requiring them to be validated on an ongoing basis. Different approaches have been used for validating these devices. Studies using treadmills shows good step-counting accuracy although test performed in field tests settings are limited. RESEARCH QUESTION: Does step-counting validity differ between a field test and a treadmill protocol for different types of activity trackers? METHODS: Thirty healthy subjects participated in this study, mean age was 28.2 (± 4.33) years, body mass 78.9 (± 12.9) kg, and height 178.5 (± 9.7) cm. A treadmill protocol with three different walking speeds (2, 3 and 4 km/h) and a 982 m field test was used. During the tests, participants' feet were filmed using a waist-mounted camera. The number of steps were extracted from the video data and used for comparison with four different step counters: a) Polar M200; b) Polar A300; c) Dunlop pedometer; d) Samsung Galaxy S9 smartphone. Validity and agreement determined was determined with the use of Bland-Altman plot and Spearman's correlation. RESULTS: Validity was higher for the field test compared to the 4 km/h treadmill test for all tested devices. The smartphone was the most accurate in terms of error, validity and agreement for both the treadmill and field test. All devices performed poorly for the 2 km/h treadmill test and only the smartphone performed well at 3 km/h. SIGNIFICANCE: The results of this study show that step counting validity and error obtained during treadmill walking is not similar to a field test. Future validation studies of activity trackers should consider this when designing a protocol. The smartphone had the lowest mean bias during the field test.