Test-retest reliability of a smartphone app for measuring core stability for two dynamic exercises

Background Recently, there has been growing interest in using smartphone applications to assess gait speed and quantify isometric core stability exercise intensity. The purpose of this study was to investigate the between-session reliability and minimal detectable change of a smartphone app for two dynamic exercise tests of the lumbopelvic complex. Methods Thirty-three healthy young and active students (age: 22.3 ± 5.9 years, body weight: 66.9 ± 11.3 kg, height: 167.8 ± 10.3 cm) participated in this study. Intraclass correlation coefficient (ICC), coefficient of variation (%CV), and Bland–Altman plots were used to verify the reliability of the test. The standard error of measurement (SEM) and the minimum detectable difference (MDD) were calculated for clinical applicability. Results The ICCs ranged from 0.73 to 0.96, with low variation (0.9% to 4.8%) between days of assessments. The Bland–Altman plots and one-sample t-tests (p > 0.05) indicated that no dynamic exercise tests changed systematically. Our analyses showed that SEM 0.6 to 1.5 mm/s-2) and MDD (2.1 to 3.5 mm/s-2). Conclusion The OCTOcore app is a reliable tool to assess core stability for two dynamic exercises. A minimal change of 3.5 mm/s-2 is needed to be confident that the change is not a measurement error between two sessions.


INTRODUCTION
Core stability has been defined as ''the capacity of the stabilizing system to maintain the intervertebral neutral zones within physiological limitations'' (Panjabi, 1992). Core stability is essential to maintain the integrity of the spinal column, provide resistance to perturbations, and supply a stable base for movement of the extremities (Gilmer et al., 2019;Panjabi, 1992).
Therapeutic exercises of the lumbopelvic complex are commonly prescribed by coaches and therapists to improve strength and facilitate more favourable lower extremity movement patterns (Bruno, 2014). Prone, quadruped, and bilateral bridge exercises generally produce low or moderate load (Ebert et al., 2017). Unilateral stance exercises in the presence of contralateral limb movement are often high or very high load activities, whilst high variability exists across a range of functional weight bearing exercises (Ebert et al., 2017).
Several studies have demonstrated a link between musculoskeletal disorders, pain and the ability to adequately control movements and muscular activation in clinical tests (Hodges & Richardson, 1996;Luomajoki et al., 2008;Moseley & Hodges, 2006). Dynamic movement tests are gaining popularity as components of musculoskeletal screening with the goal of identifying increased injury risk (Granstrom, Ang & Rasmussen-Barr, 2017). According to Barbado et al. (2016) perturbations of trunk movement can vary in amplitude and can range from self-imposed and predictable to externally imposed and unpredictable. Athletes often multitask, divide their attention, and deal with complex audiovisual spatial integration processing while engaged in physical performances (Millikan et al., 2018). Typical laboratory testing often uses sophisticated equipment, such as 3D tracking systems, force plates, and timing gates (Manor et al., 2018). Although these devices are highly valid and reliable, their expense and size make them unaffordable for use outside of a research laboratory. Accelerometers are more accessible because they are cheaper and more portable (Del Rosario, Redmond & Lovell, 2015). They are practical for measuring a long period of movements, complicated movements, and movements outdoors or over varied terrain (Del Rosario, Redmond & Lovell, 2015). Most mobile technology today comes installed with a three-dimensional accelerometer, gyroscope, and a compass with sensitivity comparable to that of research-grade biomechanical equipment (Capecci et al., 2016;Rodriguez-Sanz et al., 2018). Using a smartphone as a testing device for movement velocity has become an appealing option for researchers, coaches, and clinicians (Bilney, Morris & Webster, 2003;Cruvinel-Cabral et al., 2018;Silsupadol, Teja & Lugade, 2017). Recently, there has been growing interest in using smartphone applications to assess gait speed and quantify isometric core stability exercise intensity (Barbado et al., 2018;Silsupadol, Teja & Lugade, 2017).
However, the literature lacks studies on the reliability of smartphone apps for measuring core stability. We hypothesize that the OctoCore application is a reliable tool for measuring core stability. Thus, the objective of the present study was to investigate the between-session reliability and minimal detectable change of the OCTOcore app for two dynamic exercise tests of the lumbopelvic complex.

Study design
A descriptive repeated-measures study was performed between April and June 2018.

Participants
Thirty-three active students (12 males and 21 females; age: 22.3 ± 5.9 years, body weight: 66.9 ± 11.3 kg, height: 167.8 ± 10. three cm) were recruited from the academic community to participate in this study. Exclusion criteria were (1) any cardiovascular, respiratory, abdominal, neurological, musculoskeletal, or other chronic disease and (2) any symptoms that could affect the musculoskeletal system. Before starting the procedure, all the participants read and signed an institutional informed consent. This research project was conducted according to the Declaration of Helsinki and was approved by the CyMO Research Institute granted Ethical approval to carry out the study (1.200.518).

Procedures
The participants completed two testing sessions with intervals of 48 h. In each testing session, the participants performed two exercises tests (partial range single leg deadlift [SLD] and variations of the bird-dog exercise [BD]). These were randomized between participants, with five minutes of rest between each exercise. This is a methodological study carried out to test the reliability of the OCTOcore app. An iPhone R model 6 (iPhone R is a trademark of Apple Inc., Cupertino, CA, USA) was utilized. The iPhone R app OCTOcore was used to collect data. The mobile phone was placed, through a belt, on the midline of the subject's back at the level of the iliac crests at the level of the fourth lumbar vertebra. The headphones (EarPods; Apple Inc., Cupertino, CA, USA) were utilized to improve the concentration of the participants. According to Adusumilli et al. (2017) sampling rate was set constant at 100 Hz for both tests. The first and second repetition of each trial were discarded, analyzing a 3 s window for each repetition. The mean acceleration was calculated as the average of the acceleration magnitude data series. The acceleration data were analysed with subtraction of the gravitational component. The iPhone incorporates the Sensortec BMA280 3-axis accelerometer (Bosch Sensortec GmbH, Germany) with resolution (in ±2g range) of 0.244 mg and digital resolution of 14 bit. Several studies showed that iPhone's accelerometer is valid and reliable tool for measuring human movements (Furness et al., 2018;Nolan, Mitchell & Doyle-Baker, 2014;Sun, Wang & Banda, 2014).

Partial range single leg deadlift (SLD)
Subjects were placed in a standing position with their backs to the wall. Next, they were located at a distance of two feet from the wall, with their feet straight and spaced to the width of their hips. Also, their arms were placed crossed on their chests, and they were instructed to look forward at all times. Once a subject was placed in the starting position, a mark was placed on the ground so that the subject had a reference for where it should be placed. According to the order given by the mobile app, ''left'' or ''right'', the subject touched the right or left heel to the wall, keeping the trunk and leg straight while slightly tilting the trunk forward (Fig. 1). Then, the subject would return to the starting position to wait for the next order. This exercise provides an indication of the ability to simultaneously flex and extend at the hip with extended knees while maintaining neutral spinal alignment (Reid et al., 2015). This type of exercise produced a high or very high load (Ebert et al., 2017).

Variation of bird-dog (BD)
In the ''bird-dog'' or quadruped exercise, contralateral upper and lower extremities are lifted to horizontal from the quadruped position. The exercise was performed under the instruction that trunk motion was to be maintained to a minimum while keeping the lumbar spine and pelvis in a ''neutral'' position, knees bent to 90 • , the toes faced forward, and the hands at the participant's sides on the OctoBalance R line (Check your MOtion R , Albacete, Spain). Participants also learned to execute repetitions of the bird-dog exercise at the selected cadence, following the ticks emitted by OCTOcore app. According to the random order given by the app, ''grey left'' or ''green right,'' participants would raise their right or left arms to the side following the direction of the OctoBalance R line while stretching the opposite leg with ankle dorsi-flexion (Fig. 2). This is a conventional core stabilisation exercise that generally produced low or moderate load (Ebert et al., 2017). Each exercise (test) was performed for thirty repetitions as a familiarization trial and, after a break of three minutes, again for fifty repetitions as the measurement trial. Participants were instructed to perform the exercise at a moderate movement velocity, three seconds for each repetition per condition. All testing was performed at the same time of day to minimize the effect of circadian rhythms.

Statistical procedure
The Shapiro-Wilk test was used to check the normality of the data. Descriptive statistics were expressed as mean and standard deviation (SD). After checking the assumptions of parametric statistics, the dependent variables of the OCTOcore app (right side, left side, and composite accelerations) for two movement control tests on the different assessment days were compared by one-sample t -tests conducted for the mean difference between the two , where SD corresponds to the standard deviation from day two (Atkinson & Nevill, 1998). The minimum detectable difference (MDD) with 95% confidence interval was obtained with the equation: MDD = 1.96× √ (2×SEM) (Popovic & Thomas, 2017). All analyses assumed statistical significance at p ≤ 0.05. Statistical procedures were performed with SPSS R software version 23.0. Table 1 shows the results of the accelerometry (mm/s-2) in the two tests, in the two sessions of assessment. T -tests, ICCs, d, CV%, SEM and MDD between the first and second assessments are presented on Table 1. High ICC values (0.73-0.96) and low CV's were (<5%) observed for all exercises. Figure 3 shows Bland-Altman plots.

DISCUSSION
The aim of this study was to determine the reliability of a smartphone app (OCTOcore) for measuring core stability. The results indicate that the OCTOcore app is a reliable tool to assess core stability for two dynamic exercises.  The present study used the most popular methods to assess the reliability of medical instruments (Zaki et al., 2013).
The findings suggest this app could be used to measure the effects of an exercise intervention in in active young adults.
High ICC values were found between assessment days. Bland-Altman plots were used to visually assess the mean differences and 95% limits of agreement, showing a bias close to zero for most of the participants. The magnitude of the differences between the Day 1 and Day 2 values was small, with Cohen d values of ≤ 0.3.
In the present study, the inter-day SEM was from 0.6 to 1. six mm/s-2. This indicates the absolute consistency of the measurement. The MDD was from 2.1 to 3. five mm/s-2, which indicates that this is the smallest difference between two measurements made at different times that can be interpreted as a genuine change.
Considering that this is the first study to assess the reliability of the OCTOcore app when measuring core stability, it is difficult to draw similarities between our results and previous research. However, Barbado et al. (2018) showed that smartphone accelerometers seem to be reliable devices with which to quantify isometric core stability exercise intensity. Their data showed that most isometric control of lumbopelvic complex exercise variations (frontal bridge, back bridge, lateral bridge and bird-dog) obtained moderate-to-high reliability scores for pelvic acceleration ICC (from 0.71 to 0.88), similar to ours (from 0.73 to 0.96). In addition, Hsieh et al. (2019) recently noted evidence that a smartphone accelerometer is a valid measure of postural stability and capable of distinguishing fall risk stratification in older adults.
Moreover, providing smartphone accelerometry for control of lumbopelvic complex measurement is intended for the lay public rather than skilled researchers. Providing public access to objective assessment of lumbopelvic complex motor control may increase awareness of the dynamic control of the lumbopelvic complex and identify those in need of treatment. Because dynamic control screening is seldom conducted in clinical settings, smartphones may provide quick, objective measurement in the clinic.
OCTOcore app could potentially function as a differential method between an injured limb and an asymptomatic limb. This type of comparison is necessary to evaluate rehabilitation programs, to provide indicators for the return to the activity, or to suggest biomechanical adjustments to improve performance. Thus, future studies with OCTOcore app should include healthy and symptomatic individuals to provide parameters for treatment of injured limbs.
While this study successfully tested core stability using mobile technology, there are some limitations. The subjects of the present study were asymptomatic; therefore, the generalizability of these findings is limited, and the data obtained from healthy (asymptomatic) subjects are not representative of the population with pathologies. Another limitation of our study was that we did not establish interrater reliability. However, the known accuracy of the accelerometer should reduce rater error.

CONCLUSIONS
These results verify that the test-retest reliability of the OCTOcore app in the present study was adequate.