Reliability and Validity of the SHFT Running Power Meter

SHFT device is a novel running wearable consisting of two pods connected to your smartphone issuing several running metrics based on accelerometer and gyroscope technology. The purpose of this study was to investigate the reliability and validity of the power output (PO) metric produced by the SHFT device. To assess reliability, 12 men ran on an outdoor track at 10.5 km h 1 and 12 km h 1 on two consecutive days. To assess validity, oxygen uptake (VO 2) and SHFT data from eight men and seven women were collected during incremental submaximal running tests on an indoor treadmill on one to four separate days (34 tests in total). SHFT reliability on the outdoor track was strong with coef cients of variance (CV) of 1.8% and 2.4% for 10.5 and 12 km h 1 , respectively. We observed a very strong linear relationship between PO and VO 2(r 2 = 0.54) within subjects, and a very strong linear relationship within each subject within each treadmill test (r 2 = 0.80). We conclude that SHFT provides a reliable running power estimate and that a very strong relationship between SHFT-Power and metabolic

12 km h 1 , respectively. We observed a very strong linear relationship between PO and VO 2(r 2 = 0.54) within subjects, and a very strong linear relationship within each subject within each treadmill test (r 2 = 0.80). We conclude that SHFT provides a reliable running power estimate and that a very strong relationship between SHFT-Power and metabolic rate exists, which places SHFT as one of the leading commercially available running power meters. Keywords: sports technology; sensors; wearable; running; exercise; training; testing; tness de- vice; gadget 1. Introduction Running distances of 800 m seems to be predominated by the aerobic energy sys- tem [1,2]. Aerobic running performance is determined by maximal oxygen uptake (VO2max), running economy, and the lactate threshold [3]. Following speci c types of training regimes, increases in VO2max[4–8], running economy [9–12], and velocity at lactatethreshold [13–15] are expected. Monitoring these parameters requires costly laboratory equipment and the applicability of velocity at, e.g., lactate threshold is limited by external factors such as slope, surface, and wind resistance. Power output (PO) determination during running is a classical challenge in exercise physiology [16]. As opposed to the widely applied PO measurements in cycling, there is no simple way to quantify a runner's PO, e.g., by use of an ergometer. Thus, measures of center of gravity displacement and/or estimated cost of limb movements have been applied for decades with highly varying results and no possibility for consumer usage [16]. However, recent developments in wearable sports devices have resulted in several approaches to provide runners with an easily obtainable quanti cation of PO. Currently, at least seven running power meters are commercially available: SHFT [17], Stryd [18], RunScribe [19], Garmin Running Power [20], Polar Grit X/Vantage [21], RPM2[22], and FeetMe Sport [23]. Only a few studies have investigated the reliability and validity of these devices [24–28]. Garmin Running Power and Polar Grit X/Vantage uses barometer and GPS data, while RPM2and FeetMe Sport uses pressure and motion sensors in the shoe soles to estimate running power. SHFT, RunScribe, and Stryd all use accelerometers attached to the shoe to estimate running power,

Sport [23]. Only a few studies have investigated the reliability and validity of these devices [24–28]. Garmin Running Power and Polar Grit X/Vantage uses barometer and GPS data, while RPM2and FeetMe Sport uses pressure and motion sensors in the shoe soles to estimate running power. SHFT, RunScribe, and Stryd all use accelerometers attached to the shoe to estimate running power, and of these three, SHFT is the only device not yet scienti cally investigated. The SHFT device (Figure) Sensors2021,21, 7516.

Sensors2021,21, 7516 2 of 9 consists of two pods using accelerometer and gyroscope technology [29]. One is attached to the shoe and the other is attached to a chest band. The pods are paired with a smartphone using the of cial SHFT application. The device takes in 8000 readings per second and has an output of more than 10 metrics [17].Sensors 2021, 21, x FOR PEER REVIEW 2 of 9 power, and of these three, SHFT is the only device not yet scientifically investigated. The SHFT device (Figure 1) consists of two pods using accelerometer and gyroscope technology [29]. One is attached to the shoe and the other is attached to a chest band. The pods are paired with a smartphone using the official SHFT application. The device takes in 8000 readings per second and has an output of more than 10 metrics [17]. Figure 1. Components of the SHFT Device: A chest band, two pods, and a USB charger. Reliability is usually investigated as the coefficient of variance in a test-retest approach, whereas validity is investigated as concurrent validity by correlating PO with oxygen uptake (VO2). Of the mentioned devices Stryd is the most investigated and PO of the Stryd device appear more accurate and reliable than devices from Garmin, RunScribe, and Polar [24] with a coefficient of variance of <5% [24,25]. Regarding validity, mixed results of the Stryd device exist with coefficients of determination (PO vs. VO2) ranging from 0.08 to 0.84 [24,26,28] and 0.36 for PO vs running economy [27]. Provision of a reliable and valid running PO holds high potential for giving immediate feedback on all levels of running intensity from slow efforts to brief sprinting and intermittent exercise which is not possible with the otherwise valuable heart rate measurements. Ideally, measured running PO should accurately reflect changes in external factors such as surface, slope, and wind resistance. Additionally, accurately determined running PO has a huge potential in optimizing training and racing for athletes as well as amateurs and recreational runners. However, it can be argued that a reliable measure even with low validity is of

valuable heart rate measurements. Ideally, measured running PO should accurately reflect changes in external factors such as surface, slope, and wind resistance. Additionally, accurately determined running PO has a huge potential in optimizing training and racing for athletes as well as amateurs and recreational runners. However, it can be argued that a reliable measure even with low validity is of interest for runners since it provides the opportunity to monitor individual progress. Figure 1.Components of the SHFT Device: A chest band, two pods, and a USB charger. Reliability is usually investigated as the coef cient of variance in a test-retest approach, whereas validity is investigated as concurrent validity by correlating PO with oxygen uptake (VO2). Of the mentioned devices Stryd is the most investigated and PO of the Stryd device appear more accurate and reliable than devices from Garmin, RunScribe, and Polar [24] with a coef cient of variance of <5% [24,25]. Regarding validity, mixed results of the Stryd device exist with coef cients of determination (PO vs. VO2) ranging from 0.08 to 0.84 [24,26,28] and 0.36 for PO vs running economy [27]. Provision of a reliable and valid running PO holds high potential for giving immediate feedback on all levels of running intensity from slow efforts to brief sprinting and intermit- tent exercise which is not possible with the otherwise valuable heart rate measurements. Ideally, measured running PO should accurately re ect changes in external factors such as surface, slope, and wind resistance. Additionally, accurately determined running PO has a huge potential in optimizing training and racing for athletes as well as amateurs and recre- ational runners. However, it can be argued that a reliable measure even with low validity is of interest for runners since it provides the opportunity to monitor individual progress. The purpose of the present paper is to evaluate reliability and validity of the SHFT device PO estimate and secondary to evaluate reliability of measures that must be assumed to be reliably detected and of value for technical running analysis.

opportunity to monitor individual progress. The purpose of the present paper is to evaluate reliability and validity of the SHFT device PO estimate and secondary to evaluate reliability of measures that must be assumed to be reliably detected and of value for technical running analysis.

Sensors2021,21, 7516 3 of 9 2. Materials and Methods Data were collected on two occasions: (1) during submaximal running on an open 400 m track on two consecutive days and (2) during submaximal running on an indoor treadmill with simultaneous measurement of pulmonary gas exchange at xed velocities. 2.1. Outdoor Track Running Twelve men (35.3 11.3 years, 74.1 8.0 kg, 178.8 6.9 cm) participated in the outdoor track measurements aiming to evaluate the reliability of SHFT-sensors at two sub- maximal speeds. The SHFT device estimates several running variables (Table); however, the primary variable of interest in the current study was the estimation of PO. At two consecutive days, the subjects arrived at the same 400 m outdoor track, which conform to the standards of the International Association of Athletics Federations [30]. Subjects were asked to weigh themselves before attending and to enter their weight and height in their individual user pro le in the SHFT application on their own smartphone. At arrival, the subjects were equipped with one SHFT sensor at the bottom lace of their right foot and one SHFT sensor attached to a chest band, which were connected to their smartphone. Each subject used the same sensors across test days. The subjects warmed up for 5–10 min before running 8 min at ~10.5 km h 1 in a single le behind a pacer. After two minutes of rest, the subjects ran for 8 min at ~12 km h 1 in the same manner. A national elite runner controlled the pace using a GPS-watch as well as lap timing. The estimated PO was calculated as a 2-min average when the GPS pace was visually steady and lap timing was closest to 137 and 120 s for pace 10.5 km h 1 and 12 km h 1 , respectively. One result was excluded from the 10.5 km h 1 retest due to the chest band of a subject sliding down towards the waist. Table 1.Reliability of SHFT metrics. Indoor Treadmill Speed 1 (n = 11) Speed 2 (n = 11) Mean CV (%) Mean CV (%) Power (W) 188.6 4.6

h 1 and 12 km h 1 , respectively. One result was excluded from the 10.5 km h 1 retest due to the chest band of a subject sliding down towards the waist. Table 1.Reliability of SHFT metrics. Indoor Treadmill Speed 1 (n = 11) Speed 2 (n = 11) Mean CV (%) Mean CV (%) Power (W) 188.6 4.6 199.6 5.1 Outdoor Track 10.5 km/h (n = 11) 12 km/h (n = 12) Power (W) 191.3 1.8 207.9 2.4 Stride Rate (Stride min 1 ) 163.1 2.2 166.8 1.0 Step Length (cm) 107.7 2.8 120.9 1.7 Landing (G) 9.7 22.3 9.7 20.1 Landing Angle ( ) 19.5 20.0 21.2 15.4 Landing Position (Num) 5.3 17.1 5.7 12.6 Toe Off Angle ( ) 47.4 9.2 49.0 14.5 Contact Time (ms) 303.2 3.1 290.1 4.5 Time in Air (ms) 420.4 4.1 427.8 3.9 Deceleration (G) 13.2 23.1 16.9 15.1 Body Bounce (cm) 6.2 1.8 6.2 1.2 Running Ef ciency (%) 25.1 8.2 25.1 4.0 Running Effect (W) 47.7 6.8 51.7 5.9 Pace (km h 1 ) 10.7 2.2 12.2 2.1 Mean values for power in watts (W) and the coef cient of variation (CV) for indoor treadmill and outdoor track running tests. For the outdoor track, mean values for the different SHFT metrics from the two testing days are also provided. 2.2. Indoor Treadmill Eight men and seven women (26 3 years, 66.3 9.0 kg, 176 10 cm and a maximal oxygen uptake of 57 9 mL/kg/min) completed the indoor treadmill measurements 1–4 timesseparated by 1–5 weeks. Participants were tested on a treadmill (The Pro, Wood- way USA, Inc., Waukesha, WI, USA) as illustrated in Figures, starting with a 10 min warmup at the same absolute velocity across all test days, which was individualized for each subject. After the warm-up, the subjects were equipped with a SHFT-sensor (SHFT, Copenhagen, Denmark) on a random lace of the left or right foot and a SHFT-sensor at-

which was individualized for each subject. After the warm-up, the subjects were equipped with a SHFT-sensor (SHFT, Copenhagen, Denmark) on a random lace of the left or right foot and a SHFT-sensor at-

Sensors2021,21, 7516 4 of 9 tached to a chest band, which were connected to a smartphone using the SHFT application. Additionally, subjects were equipped with a mask connected to a mixing chamber for mea- suring pulmonary gas exchange of O2and CO2using an automated metabolic gas analysis system (Quark CPET, COSMED, Rome, Italy). Five minutes after the warm-up, the subjects initiated a submaximal running test with 3–5 speed increases of 1 km h 1 every 3 min. The speed increase continued until capillary blood lactate values were >4 mmol L 1 , which was measured in the nal minute of each speed level using an ABL 800 Flex (Radiometer, Brønshøj, Denmark). The coef cient of variance (CV) was calculated for all subjects who completed the indoor running test at least twice (n = 11) at two submaximal velocities. The coef cient of determination (r 2 ) was calculated using all 34 tests for all subjects (n = 15). Oxygen uptake and SHFT data were calculated as an average of 30 s between time 1:10 and 1:40 at each speed.Sensors 2021, 21, x FOR PEER REVIEW 4 of 9 2.2. Indoor Treadmill Eight men and seven women (26 ± 3 years, 66.3 ± 9.0 kg, 176 ± 10 cm and a maximal oxygen uptake of 57 ± 9 mL/kg/min) completed the indoor treadmill measurements 1–4 times separated by 1–5 weeks. Participants were tested on a treadmill (The Pro, Woodway USA, Inc., Waukesha, WI, USA) as illustrated in Figures 2 and 3, starting with a 10 min warmup at the same absolute velocity across all test days, which was individualized for each subject. After the warm-up, the subjects were equipped with a SHFT-sensor (SHFT, Copenhagen, Denmark) on a random lace of the left or right foot and a SHFT-sensor attached to a chest band, which were connected to a smartphone using the SHFT application. Additionally, subjects were equipped with a mask connected to a mixing chamber for measuring pulmonary gas exchange of O2 and CO2 using an automated metabolic gas analysis system (Quark CPET, COSMED, Rome, Italy). Five minutes after the warm-up, the

or right foot and a SHFT-sensor attached to a chest band, which were connected to a smartphone using the SHFT application. Additionally, subjects were equipped with a mask connected to a mixing chamber for measuring pulmonary gas exchange of O2 and CO2 using an automated metabolic gas analysis system (Quark CPET, COSMED, Rome, Italy). Five minutes after the warm-up, the subjects initiated a submaximal running test with 3–5 speed increases of 1 km·h −1 every 3 min. The speed increase continued until capillary blood lactate values were >4 mmol·L −1, which was measured in the final minute of each speed level using an ABL 800 Flex (Radiometer, Brønshøj, Denmark). The coefficient of variance (CV) was calculated for all subjects who completed the indoor running test at least twice (n = 11) at two submaximal velocities. The coefficient of determination (r 2) was calculated using all 34 tests for all subjects (n = 15). Oxygen uptake and SHFT data were calculated as an average of 30 s between time 1:10 and 1:40 at each speed. Figure 2. Schematic of the workflow in the indoor treadmill test. Solid boxes indicate speeds which were completed in all 34 tests, whereas speed 4 and speed 5 were completed in 24 tests and 1 test, respectively. Arrows indicate capillary blood sampling. Figure 3. Demonstration of the indoor treadmill test setup where (A) illustrates a SHFT pod equipped on a random lace, (B) illustrates a SHFT pod equipped on a chest band, and (C) illustrates the mixing chamber positioned on a metabolic cart. Time (min): –15 –5 Warm-up 0 3 6 9 12 15 Equipment Phase Speed 1 Speed 2 Speed 3 Speed 4 Speed 5 Capillarybloodsample SHFT and VO 2 sampling Intensity Figure 2. Schematic of the work ow in the indoor treadmill test. Solid boxes indicate speeds which were completed in all 34 tests, whereas speed 4 and speed 5 were completed in 24 tests and 1 test, respectively. Arrows indicate capillary blood sampling.Sensors 2021, 21, x FOR PEER REVIEW 4 of 9 2.2. Indoor Treadmill Eight men and seven women (26 ± 3

of the work ow in the indoor treadmill test. Solid boxes indicate speeds which were completed in all 34 tests, whereas speed 4 and speed 5 were completed in 24 tests and 1 test, respectively. Arrows indicate capillary blood sampling.Sensors 2021, 21, x FOR PEER REVIEW 4 of 9 2.2. Indoor Treadmill Eight men and seven women (26 ± 3 years, 66.3 ± 9.0 kg, 176 ± 10 cm and a maximal oxygen uptake of 57 ± 9 mL/kg/min) completed the indoor treadmill measurements 1–4 times separated by 1–5 weeks. Participants were tested on a treadmill (The Pro, Woodway USA, Inc., Waukesha, WI, USA) as illustrated in Figures 2 and 3, starting with a 10 min warmup at the same absolute velocity across all test days, which was individualized for each subject. After the warm-up, the subjects were equipped with a SHFT-sensor (SHFT, Copenhagen, Denmark) on a random lace of the left or right foot and a SHFT-sensor attached to a chest band, which were connected to a smartphone using the SHFT application. Additionally, subjects were equipped with a mask connected to a mixing chamber for measuring pulmonary gas exchange of O2 and CO2 using an automated metabolic gas analysis system (Quark CPET, COSMED, Rome, Italy). Five minutes after the warm-up, the subjects initiated a submaximal running test with 3–5 speed increases of 1 km·h −1 every 3 min. The speed increase continued until capillary blood lactate values were >4 mmol·L −1, which was measured in the final minute of each speed level using an ABL 800 Flex (Radiometer, Brønshøj, Denmark). The coefficient of variance (CV) was calculated for all subjects who completed the indoor running test at least twice (n = 11) at two submaximal velocities. The coefficient of determination (r 2) was calculated using all 34 tests for all subjects (n = 15). Oxygen uptake and SHFT data were calculated as an average of 30 s between time 1:10 and 1:40 at each speed. Figure 2. Schematic of the workflow in the indoor treadmill test. Solid boxes indicate speeds which were completed in all 34 tests, whereas speed

This study evaluates the reliability and validity of the SHFT running power meter.