Validity of the iLOAD® app for resistance training monitoring

Participants

Sixteen young healthy individuals, four women (mean ± standard deviation [SD]; age = 29.5 ± 7.2 years, body height = 1.61 ± 0.07 m, body mass = 58.7 ± 6.1 kg, squat one repetition maximum [1RM] = 53.0 ± 18.8 kg) and 12 men (mean ± SD; age = 27.4 ± 7.2 years, body height = 1.76 ± 0.05 m, body mass = 78.7 ± 8.2 kg, squat 1RM = 102.7 ± 15.4 kg), volunteered to participate in this study. All participants were familiar with the half squat exercise and had at least one year of RT experience. Prior to testing and after detailed explanation of the procedures and risks of the study, participants gave their written consent to participate in the study. The study protocol adhered to the tenets of the Declaration of Helsinki and was approved by the Catholic University of Brasilia (54813016.0.0000.0029).

Study design

This study was designed to determine the validity of the iLOAD^® app for monitoring MV and total work during a RT session (Fig. 1). Participants came to the laboratory on two occasions separated by 48 h. The 10RM load was determined in the first testing session during the half squat exercise. In the second testing session participants were instructed to perform three sets of 10 repetitions during the half squat exercise against the 10RM load. Participants completed 9.9 ± 0.5, 9.5 ± 1.3, and 9.7 ± 1.3 repetitions during the first, second and third set, respectively. Both the iLOAD^® app (v 1.0; ILoad Solutions, Brasilia, Brazil) and a linear encoder (Chronojump, Barcelona, Spain) were used to calculate MV and total work of each training set. After familiarization with the iLOAD^® app, two independent researchers recorded MV over the training sets with the iLOAD^® app to evaluate the inter-rater agreement. The average value of both raters was considered to explore the concurrent validity of the iLOAD^® app with respect to the linear encoder. Both testing sessions were performed at the same time of the day for each participant.

Procedures

All measurements were conducted at the same laboratory (“Laboratório de Estudos da Força” of “Universidade Católica de Brasília”). Anthropometric data was assessed at the beginning of the first session with a stadiometer (ES2040; Sanny, São Paulo, Brazil) and an electronic scale (W110 H LED; Welmy, Santa Bárbara d’Oeste, Brazil). The vertical distance of the barbell for each participant between a knee angle of 90° and the standing position (hips and knees fully extended and feet flat on the floor) was determined with a measuring tape (E095; Stamaco, São Paulo, Brazil). This distance was introduced in the iLOAD^® app for the computations of MV and total work. This measurement was collected while the participants held the unloaded Smith machine barbell (17 kg). The 90° knee angle was determined with a manual goniometer (187-907; MITUTOYO^®, São Paulo, Brazil). To ensure a consistent countermovement depth during all repetitions, an elastic cord was positioned to contact with the participants’ buttocks when they reached a 90° knee angle. In addition, an adhesive tape was located on the ground to instruct the participants to place their feet always in the same position. After these measurements, participants completed a warm-up consisting of 5 min of submaximal running on a treadmill at a self-selected pace, followed by 15 repetitions with the 17 kg bar of the Smith machine used in the present study (Power Tech, Righetto, São Paulo, Brazil). Thereafter, the 10RM load during the half squat exercise was evaluated following the protocol proposed by the American College of Sports Medicine (i.e., four attempts separated with a 3-min recovery interval) (Pescatello et al., 2014). The initial load corresponded to 70% of the self-perceived 10RM, and it was progressively increased until the participant could not complete more than 10 repetitions.

The warm-up of the second testing session consisted of 5 min of submaximal running on a treadmill at a self-selected pace, 15 repetitions with the unloaded Smith machine barbell, and five repetitions with the previously determined 10RM load. Three minutes after completing the last warm-up set, participants were instructed to perform three sets of 10 repetitions during the half squat exercise against the 10RM load with 3 min of rest between sets. Participants were instructed to complete all sets as quickly as possible maintaining the same range of motion during all repetitions. All sets started with a ’go’ instruction from one of the raters, and were considered finished when the concentric phase of the last repetition was completed.

Data acquisition and analysis

Mean velocity (MV) and total work of the three sets were calculated from the recordings of both the iLOAD^® app and the linear encoder:

- iLOAD^® app: The iLOAD^® app was installed in two smartphones (5S, iPhone, USA) with an actualized operating system (11.2, IOS, USA). Two independent raters were positioned in front of the participants and recorded the time needed to complete each set. The raters were familiarized with procedures during two preliminary sessions which consisted of the same protocols. The inputs of the iLOAD^® app for each set were the load (in kilograms), number of repetitions, vertical distance of the barbell, and time needed to complete the set (in seconds). The time needed to complete each set was determined by the chronometer of the smartphone. The smartphone’s chronometer was initiated after the ’go’ signal used to indicate the start of the set in the iLOAD^® app and it was stopped when the subject completed the last repetition of the set (i.e., when the hips and knees reached full extension). Of note, because of the greater mechanochemical efficiency during the eccentric action (i.e., negative work) when compared to the concentric action (i.e., positive work) of a single complete repetition (De Looze et al., 1994; Ryschon et al., 1997), a factor of 1.33 rather than 2 was considered for the sum of the concentric and eccentric phases for total work calculations (Bloomer et al., 2006). In addition, the iLOAD^® app was used in the ‘squatting’ mode that considers the user’s body mass with a weighting factor of 0.88 (Bloomer et al., 2006) for computing total work. The MV and total work of each set were automatically calculated by the iLOAD^® app as follows: (1)${\displaystyle Meanvelocity\left(m∕s\right)=\frac{Numberofrepetitions\times 2\times Verticaldistance\left(m\right)}{Timeneededtocompletetheset\left(s\right)}}$ (2)${\displaystyle Totalwork\left(J\right)=1.33\times Verticaldistance\left(m\right)\times Numberofrepetitions\times \left[\left(Bodymass\times 0.88\right)+Load\left(kg\right)\right]\times 9.81\left(m∕s^2\right)}$

- Linear encoder: The linear encoder (Chronojump, Barcelona, Spain) was connected to a laptop (NP540U3C, Samsung Electronics Co., China) with custom made software (1.8.1-95, Chronojump, Barcelona, Spain). The cable of the linear encoder was fixed perpendicularly to the barbell and recorded displacement-time data at 1,000 Hz. An Excel^® spreadsheet was used to calculate the MV and total work from the raw displacement-time data provided by the software as follows: (3)${\displaystyle Meanvelocity\left(m∕s\right)=\frac{Downwarddisplacement\left(m\right)+Upwarddisplacement\left(m\right)}{Timeneededtocompletetheset\left(s\right)}}$ (4)${\displaystyle Totalwork\left(J\right)=1.33\times Verticaldistance\left(m\right)\times Numberofrepetitions\times \left[\left(Bodymass\times 0.88\right)+Load\left(kg\right)\right]\times 9.81\left(m∕s^2\right)}$

Statistical analysis

Descriptive data are presented as means and SD. Normality of the distribution was confirmed by the Shapiro–Wilk test (p > 0.05). The inter-rater agreement for the recordings of MV with the iLOAD^® app, as well as the concurrent validity of the iLOAD^® app respect to the linear encoder for measuring MV and total work were assessed by independent samples t-tests, Cohen’s d effect size (ES and 95% confidence interval [CI]), Pearson’s correlation coefficients (r) and Bland-Altman plots. Note that the total work was not compared between raters because its value does not depend on the time recorded by the raters (see Eq. (2)). The scales proposed by Hopkins et al. (2009) were used to interpret the magnitude of the ES (trivial <0.20, small = 0.20–0.60, moderate >0.60–1.20, large >1.20–2.00, and extremely large >2.00) and r coefficients (trivial <0.10, small = 0.10–0.30, moderate >0.30–0.50, high >0.50–0.70, very high >0.70–0.90, or practically perfect >0.90). Heteroscedasticity of error was defined as a r² > 0.10 (Atkinson & Nevill, 1998). All statistical analyses were performed using SPSS software version 22.0 (SPSS Inc., Chicago, IL, USA) and statistical significance was set at an alpha level of 0.05.

Inter-rater agreement

No significant differences and nearly perfect correlations were observed between raters for the MV values collected under individual sets (p ≥ 0.38, ES ≤ 0.02, r ≥ 0.987) as well as for the whole training session (p = 0.38, ES = 0.01, r = 0.997). Bland-Altman plots also revealed very low systematic bias (≤0.003 m s⁻¹) and random errors (≤0.033 m s⁻¹), while heteroscedasticity of the errors was observed for the sets 2 (r² = 0.208) and 3 (r² = 0.199), but not for the set 1 (r² = 0.045) or the whole training session (r² = 0.074) (Fig. 2).

iLOAD^® app vs. linear encoder

Due to the very high inter-rater agreement reported above, the average value of both raters was considered to explore the concurrent validity of the iLOAD^® app with respect to the linear encoder. Although significant differences between the iLOAD^® app and the linear encoder were observed for MV during the sets 2 and 3 as well as for the whole training session (p < 0.05), the magnitude of the differences were trivial-small and the correlations were always nearly perfect between both devices for MV and total work (Table 1). Bland-Altman plots did not reveal heteroscedasticity of the errors between the iLOAD^® app and the linear encoder for MV and total work (r² ≤ 0.010) (Fig. 3).