Maximum velocity during loaded countermovement jumps obtained with an accelerometer, linear encoder and force platform: A comparison of technologies

1 TITLE 1 Maximum velocity during loaded countermovement jumps obtained with an 2 accelerometer, linear encoder and force platform: a comparison of technologies 3 4 5 6 2 ABSTRACT 1 The maximum velocity (Vmax) reached during countermovement jumps (CMJ) 2 has been considered a performance indicator to evaluate the vertical jump ability. The 3 aim of this study was to compare Vmax during loaded CMJ (CMJload) using three 4 different technologies to show a criterion for to select the more appropriate depending 5 of the use. Nine recreationally active men performed a CMJload test. Five jumps were 6 made in each of 6 series with a 20-kg barbell + 0, + 5, + 10, + 15, + 20 and + 25 kg, 7 with 2 seconds of rest between the jumps and 5 minutes of rest between the series to 8 explore a wide range of speeds. Vmax was obtained from force platform, inertial device 9 and linear encoder technologies. To compare devices, Bland-Altman plots and mean 10 differences were used. The reproducibility was tested using the intraclass correlation 11 coefficient (ICC) for single measures and typical error (TE). All technologies showed 12 high levels of reproducibility, ICC higher than 0.75 and TE lower than 10%. There were 13 non-significant differences in Vmax between each pair of technologies (linear encoder 14 2.11 ± 0.24 m·s-1, inertial device 2.11 ± 0.26 m·s-1, force platform 2.13 ± 0.25 m·s-1) 15 reporting a very low bias. Although the limits of agreement between the different 16 technologies evaluated were high (± 0.33 m·s-1). In conclusion, the accelerometer, linear 17 encoder and force platform were suitably reliable to be used to measure the Vmax 18 during loaded vertical jumps but theirs values were not interchangeable. 19 20 3 1. Introduction 1 Resistance training is considered a valuable method to improve sports 2 performance. Coaches and strength and conditioning professionals use the vertical force 3 application as a key factor to success in competitions such as team sports or athletics 4 (González-Badillo and Marques, 2010). Hence, a typical exercise to improve the 5 vertical applied force during the resistance training is the loaded countermovement 6 jump (CMJ-loaded) (Jiménez-Reyes et al., 2016). Jump height has been considered an 7 indicator of the vertical force applied during the CMJ jump (Samozino et al., 2010). 8 Most devices use flight time to calculate the jump height (Buckthorpe et al., 2012; 9 McMahon et al., 2016). Any difference between the athlete's take-off and landing 10 position could increase the flight time but not the jump height. The center of mass 11 (COM) height in the take-off and landing must coincide for a correct estimate of jump 12 height (Dias et al., 2011; Picerno et al., 2011). Recently, it has been shown that to 13 consider the COM height equal as the real jump height is a mistake (R3). Considering 14 this technique implication, take-off velocity has been suggested as a more proper 15 parameter for assessing jump height (Jiménez-Reyes et al., 2016). This is calculated 16 using the impulse-momentum method from records of a force platform (García-Ramos 17 et al., 2015; Glatthorn et al., 2011; Jiménez-Reyes et al., 2016; Taylor et al., 2010). 18 Nevertheless, the impulse-momentum method has been shown to be more precisely 19 compared to other calculation methods using force platforms (R3). Due to the difficulty 20 in determining the precise instant of take-off, the Vmax achieved during the upward 21 movement has been proposed as another option (García-Ramos et al., 2015; Jiménez- 22 Reyes et al., 2016; Taylor et al., 2010). Although identifying the instant in which the 23 Vmax is reached could be clearer than the take-off instant, few studies have used this 24 4 variable to evaluate the vertical jump performance. Consequently, more studies 1 exploring the utility of the Vmax are needed. 2 Recently, accelerometers have also been used to calculate the jump height 3 (Casartelli et al., 2010; Magnúsdóttir et al., 2014; Picerno et al., 2011). These systems 4 measure the acceleration related to gravity (g). Thus, when the accelerometer registers 1 5 g, it can be assumed that the athlete is in flight (Casartelli et al., 2010; Magnúsdóttir et 6 al., 2014; Picerno et al., 2011). What is more, it is possible to calculate both the time the 7 athlete is in flight as the velocity at take-off obtained from the single numerical 8 integration of vertical acceleration. However, no studies have been found which use the 9 Vmax calculated by accelerometers to evaluate the vertical jump performance. 10 The objectives of this study were firstly to compare Vmax obtained with an 11 accelerometer, a linear encoder and a force platform executing CMJ-load; secondly, to 12 determine the reliability of three devices to estimate the Vmax. For the first objective, it 13 was hypothesized that there would be high agreement in the Vmax estimation between 14 the three devices. For the second objective, it was hypothesized that the degree of 15 reliability to calculate the Vmax of the three devices was similar. 16 2. Methods 17 2.1 Subjects 18 A total of nine recreationally active, healthy sports science students (N=9, 19 age=20.78 ± 2.11 years, height=1.76 ± 0.05 m, body mass=74.50 ± 10.55 kg) were 20 recruited for this study. None of the participants were experiencing injuries that could 21 have potentially constrained their ability to perform the research task. The University 22 Ethics Committee approved the study and it was undertaken according to the Helsinki 23 Conference for research on humans (Williams, 2008). The volunteer participants were 24 5 given full information about the purpose of the study and gave their informed consent 1 before participation. 2 2.2 Experimental Procedures 3 The participants completed a standard 10-min warm-up composed of biking, 4 lower body dynamic stretches, deep squats and vertical jumps. Then, each participant 5 performed an incremental load CMJ test on a Smith machine (Multipower Fitness Line, 6 Peroga, Spain) to allow only vertical displacement of the barbell during CMJload. 7 Previous studies recommended the use of Smith machine to determine vertical jump 8 height during the Squat-load jump (Pérez Castilla et al. 2017). Low, moderate and high 9 loads were chosen to compare between devices ensuring a wide range of speed 10 achieved. Five jumps were made in each of 6 series with a 20-kg barbell + 0, + 5, + 10, 11 + 15, + 20 and + 25 kg, with two seconds of rest between jumps and 5 minutes of rest 12 between series. A force platform (Kistler Instruments Ltd., Hook, UK) was placed on 13 the floor over a horizontal steel plate. One multisensor WIMU containing a 3D 1000 Hz 14 accelerometer (RealTrack System, Almería, Spain) and a linear encoder (SmartCoach, 15 Stockholm, Sweden) were tightly fixed to the barbell (Figure 1). 16 17 *** Insert Figure 1 about here *** 18 2.3 Instruments 19 Force platform. Data were obtained with the software BioWare V5.3.0.7 (Kistler 20 Holding AG, Switzerland). The force platform recorded data at a sampling frequency of 21 1000 Hz. To remove random noise, all measurements were low-pass-filtered at a cut-off 22 frequency of 100 Hz using a 2nd-order Butterworth filter. This frequency was 23 previously selected according to the residual analysis results (Winter, 2009). The 24 6 vertical component of center of mass velocity was estimated using the impulse- 1 momentum method. Net impulse was obtained by integrating the net vertical ground 2 reaction force using the trapezoid method (Buckthorpe et al., 2012). In our study, the 3 force platform was selected as reference (gold standard) for further comparisons. 4 Accelerometer. An inertial device (WIMU, RealTrack Systems, Almeria, Spain) 5 which contains a 3D accelerometer was used to measure the acceleration of the barbell 6 at a sampling frequency of 1000 Hz. The calibration of the accelerometer was verified 7 previously by checking that the accelerometer supported a flat surface that was 8 constantly measured as 1 g over 5 minutes of recording. The vertical axis was 9 determined in relation to the axis of the earth to maintain the verticality of the 10 movement without affecting the lateral inclination of the barbell where the device was 11 placed. A software provided by the manufacturer (Quiko V.0.882, RealTrack System, 12 Almería, Spain) was used to obtain the vertical velocity from accelerometer data. A 13 low-pass-filter at a cut-off frequency of 10 Hz using a 2nd-order Butterworth filter was 14 used to remove random noise of the acceleration data. This frequency was previously 15 selected according to the residual analysis results (Winter, 2009). 16 Linear encoder. The linear encoder consisted of a cable extension linear velocity 17 transducer. Vertical instantaneous velocity was directly sampled by the device at a 18 frequency of 100 Hz with the software SmartCoach V4.0.12 (SmartCoach EU, 19 Stockholm, Sweden), with data filtered with a 10 point rolling average filter. 20 Previously, the linear encoder had been validated in a certified laboratory, and the 21 obtained certified quality was ISO 90019:2008 with an average error of 0.52% ± 0.17%. 22 2.4 Statistical analysis 23 7 Data are shown as the mean ± standard deviation (SD). Normal distribution was 1 checked with the Kolmogorov-Smirnov and Shapiro-Wilk normality tests, depending on 2 sample size. Vmax possible mean differences between pairs of instruments were 3 calculated via general linear model for repeated measures. In addition, Bland-Altman 4 plot analysis was performed to compare Vmax bias between each pair of instruments. 5 From these plots, bias, confidence intervals such as SD * 1.96 and the Bland-Altman 6 correlation coefficient (r2) were calculated. 7 To test the reproducibility of the data, the intraclass coefficient correlation (ICC) 8 for single measures was calculated for each device and intra-trials. Additionally, the 9 typical error (TE) as a percentage of the coefficient of variation (%CV) and as the 10 absolute value was calculated to provide information about the error in the 11 measurements. In addition, the smallest worthwhile change (SWC) was calculated as 12 0.5 x SD for each data set from each instrument. Generally, an ICC higher than 0.75 and 13 TE lower than 10% were interpreted to reflect good reliability. The level of statistical 14 significance was set at P < 0.05 for the mean comparisons and P < 0.01 for the 15 correlation analysis. Calculations were performed using SPSS Statistics 22 for Windows 16 (IBM Co., NY. USA), and an Excel spreadsheet was made for specific calculation of the 17 TE. 18 3. Results 19 Descriptive values and reliability results for Vmax are shown in Table 1. Bland- 20 Altman Plots are shown in Figure 2. The linear adjustment correlation coefficient for 21 the Bland-Altman Plots (r2) showed stable differences within the data range for each 22 pair of comparisons among the instruments (r < 0.009). The mean BIAS ± LoA for the 23 8 linear encoder and accelerometer were 0.02 ± 0.35 m·s-1 and 0.01 ± 0.34 m·s-1, 1 respectively. 2 *** Insert Table 1 about here *** 3 *** Insert Figure 2 about here *** 4 The ICCs for all the instruments were higher than 0.75 (linear encoder = 0.944 5 (0.913 to 0.966, 95% confidence intervals (CI)), accelerometer = 0.815 (0.748 to 0.882 6 95% CI), and force platform = 0.924 (0.884 to 0.953 95% CI). The TE was lower than 7 10% for all the instruments (linear encoder = 2.87%, accelerometer = 4.88%, and force 8 platform = 2.87%). The SWC values for the linear encoder, accelerometer and force 9 platform were 0.12 m·s-1, 0.13 m·s-1 and 0.12 m·s-1, respectively (Table 1). 10 4. Discussion 11 The main findings of this study confirmed our first hypothesis. No difference 12 was found in the Vmax calculated between the three devices. In addition, the three 13 devices were shown to be consistent when calculating Vmax with high levels of 14 reproducibility. However, wide LoA were found in both linear encoder and 15 accelerometer when compared to the gold standard. 16 Previous studies have investigated the validity of an accelerometer to measure 17 jump height, using mainly three methods to estimate jump height: flight time calculation 18 (Casartelli et al., 2010; Castagna et al., 2013; Choukou et al., 2014; Lesinski et al., 19 2016; Magnúsdóttir et al., 2014; Picerno et al., 2011), direct vertical acceleration 20 measurement (Picerno et al., 2011) or velocity derived calculation (Choukou et al., 21 2014). Most studies reported systematic differences in the jump height calculation 22 between the accelerometer and other devices, such as force platforms. To our 23 knowledge, in agreement with our results, only three studies have shown valid results 24 9 on jump height calculations compared with gold standards (Castagna et al., 2013; 1 Picerno et al., 2011), although they used the flight time calculation method. Despite the 2 low validity usually shown by other studies, they observed high reliability data 3 (Casartelli et al., 2010; Choukou et al., 2014; Lesinski et al., 2016; Picerno et al., 2011), 4 which is in agreement with our results. The main differences found in the validity data 5 between our study and others may be explained by the position of the accelerometer on 6 the barbell of the Smith machine. Recently, it has been shown that the position of the 7 accelerometer in the COM many not necessary reflects the real athlete jump height 8 (R3). 9 The results of present study suggested that these three devices are not 10 interchangeable to measure jump height, due to wide LoAs found in the Vmax 11 calculation. However, the reliability study showed that the SWC was close to 0.1 m·s-1 12 and superior to TE (0.06 m·s-1 to 0.09 m·s-1), which corresponded to a valid situation to 13 detect changes in performance (Sánchez-Medina and González-Badillo, 2011). In 14 addition, the reliability study showed positive values for the ICC (>0.75) and TE 15 (<10%). For this reason, any of the three instruments were able to detect performance 16 changes in these kinds of tests. This study shows results that will allow to practitioners 17 to obtain a criterion for the selection of the instruments according to the purpose of their 18 work, the usability, economy and portability. 19 A limitation of the present study was that the accelerometer was fixed to a 20 barbell located in a Smith machine ensuring a vertical path. Since it is usual in training 21 sessions to execute load jumps without the movement restrictions of a Smith machine, 22 further studies are necessary which confirm these results when the jumps are executed 23 freely for CMJ-load. Similarly, the results of this study would be limited to vertical 24 barbell jumps. Thus, more investigation is needed to test whether the accelerometer 25