The reliability of accelerometry to measure

This article was downloaded by: [Kimitake Sato] On: 08 November 2012, At: 04:57 Publisher: Routledge Informa Ltd Registered in England and Wales Registered Number: 1072954 Registered office: Mortimer House, 37-41 Mortimer Street, London W1T 3JH, UK

Sports Biomechanics Publication details, including instructions for authors and subscription information: http://www.tandfonline.com/loi/rspb20

The reliability of accelerometry to measure weightlifting performance a

a

Kimitake Sato , William A. Sands & Michael H. Stone

a

a

Center of Excellence for Sport Science and Coach Education, East Tennessee State University, Johnson City, TN, USA Version of record first published: 04 Oct 2012.

To cite this article: Kimitake Sato, William A. Sands & Michael H. Stone (2012): The reliability of accelerometry to measure weightlifting performance, Sports Biomechanics, 11:4, 524-531 To link to this article: http://dx.doi.org/10.1080/14763141.2012.724703

PLEASE SCROLL DOWN FOR ARTICLE Full terms and conditions of use: http://www.tandfonline.com/page/terms-and-conditions This article may be used for research, teaching, and private study purposes. Any substantial or systematic reproduction, redistribution, reselling, loan, sub-licensing, systematic supply, or distribution in any form to anyone is expressly forbidden. The publisher does not give any warranty express or implied or make any representation that the contents will be complete or accurate or up to date. The accuracy of any instructions, formulae, and drug doses should be independently verified with primary sources. The publisher shall not be liable for any loss, actions, claims, proceedings, demand, or costs or damages whatsoever or howsoever caused arising directly or indirectly in connection with or arising out of the use of this material.

Sports Biomechanics November 2012; 11(4): 524–531

The reliability of accelerometry to measure weightlifting performance

KIMITAKE SATO, WILLIAM A. SANDS, & MICHAEL H. STONE

Downloaded by [Kimitake Sato] at 04:57 08 November 2012

Center of Excellence for Sport Science and Coach Education, East Tennessee State University, Johnson City, TN, USA (Received 7 April 2012; accepted 15 August 2012)

Abstract The purposes of the study were to track weightlifters’ barbell acceleration with a portable accelerometer over three training sessions to examine test – retest reliability and to compare peak barbell acceleration at different training intensities. Twelve nationally ranked weightlifters volunteered for this study. The portable accelerometer was attached to the right side of the barbell to measure barbell resultant acceleration during the snatch lift at a sampling frequency of 100 Hz. The data were collected over three training sessions at intensity levels of 80%, 85%, and 90% of one repetition maximum. The data were analyzed using intra-class correlation coefficients (ICCs) for the three training sessions and oneway repeated measure ANOVA to compare the difference in peak barbell acceleration at three intensities. Results showed that the device was highly reliable with an ICC of 0.88 and 95% confidence interval of 0.81 – 0.93. There were significant differences in peak barbell acceleration at various lifting intensities, indicating a decline of the acceleration as the mass of the barbell became heavier. The portable accelerometer seems useful in measuring barbell acceleration data, which can be analyzed in future studies to monitor a weightlifter’s performance in a practical setting instead of testing at a laboratory.

Keywords: Barbell acceleration, training intensity, portable accelerometer, snatch lift

Introduction When designing a workout program, a typical procedure to estimate appropriate workload is to utilize a one repetition maximum (1RM) to calculate training volume loads (sets, repetitions, and weights). This approach has been one of the most useful procedures to improve strength from youth to elite levels (Baechle & Earle, 2000; Stone et al., 2006a, 2006b). Even though this approach may be one of the most reliable and accurate procedures in training weightlifters and other athletes, coaches are still required to have a keen eye to detect athletes’ small day-to-day performance change. If coaches do not have enough experience and the appropriate tools to detect the small changes, they may find some difficulty in detecting a level of improvement or physical fatigue. Correspondence: Kimitake Sato, Center of Excellence for Sport Science and Coach Education, East Tennessee State University, Campus Box 70654, Johnson City, TN 37614, USA, E-mail: [email protected] ISSN 1476-3141 print/ISSN 1752-6116 online q 2012 Taylor & Francis http://dx.doi.org/10.1080/14763141.2012.724703


The reliability of accelerometry

525

In order to evaluate lifting technique, motion analysis software is necessary. The software provides quick visual feedback to detect errors in techniques. Quantitative measures provide more sophisticated biomechanical data to evaluate the lifting technique and performance (Isaka et al., 1996; Lee et al., 1996; Hiskia, 1997; Stone et al., 1998; Gourgoulis et al., 2000; Schilling et al., 2002; Haff et al., 2003; Gourgoulis et al., 2004). In reality, if a lifter requests biomechanical assessment, a testing site usually requires a laboratory setting. Laboratory testing is logistically difficult for lifters and coaches because it is time consuming and the athletes’ time can be constrained by the laboratory’s schedule. Availability of sophisticated laboratory testing is also limited for many lifters. For these reasons, it could be beneficial to have easy-to-use testing instruments to capture and evaluate lifters’ performance at the training site with real-time feedback to coaches and athletes. During the last 20 years, the biomechanics of weightlifting for both the snatch and clean and jerk have been studied extensively. The intent of analyzing lifting technique has been to (1) understand how successful lifts are different from unsuccessful lifts and (2) establish a baseline of typical technique of elite weightlifters (Garhammer & Hatfield, 1985; Bartonietz, 1996; Isaka et al., 1996; Barton, 1997; Hiskia, 1997; Stone et al., 1998; Gourgoulis et al., 2000; Schilling et al., 2002; Haff et al., 2003; Gourgoulis et al., 2004; Sato et al., 2009a). Commonly measured variables from previous studies are barbell path, velocity, mechanical work, and power. However, Stone et al. (2006a) postulated that it is difficult to predict perfect technique based on biomechanical characteristics because of small inter-lifter variations in lifting technique. For example, although peak power is consistently higher among world-class lifters than at lower competitive levels, there are also some variations in kinematics such as barbell velocity and trajectory (Garhammer & Hatfield, 1985; Stone et al., 2006a). When speaking about lifting technique from an anthropometric perspective, smaller lifters may not require high barbell velocities because they require shorter vertical barbell displacement for the catch in both snatch and clean as compared to taller lifters. Rather, research focus should be examining the optimal barbell velocity in order to achieve a desired barbell height for the catch. Barbell acceleration, which is a fundamental variable for kinematic analysis, seems an under-reported variable in weightlifting studies. Studies (Garhammer & Hatfield, 1985; Gourgoulis et al., 2000) have displayed a vertical barbell acceleration graph, but no further discussion was provided regarding the interpretation of the graph. As barbell acceleration is proportional to the force applied to the barbell, barbell acceleration could be an important variable to assess weightlifting performance. A recently published article reported that a wireless accelerometer is an economical device that accurately measures acceleration as well as an expensive high-speed camera for kinematic analysis (Sato et al., 2009b). The data transfer from the device to a laptop occurs wirelessly thus avoiding any interference with the athlete from the researchers during the training session. The software allows immediate feedback, and most importantly, data collection can be done at a training site rather than lifters reporting to a laboratory. A portable accelerometer has been used in performance-based studies, and it has proven helpful in the assessment of athletes’ performances (Casatelli et al., 2010; Kraemer, 2010). Using this type of wireless accelerometer could be useful in evaluating weightlifting performance, and there is a need for testing an accelerometer in the training environment across multiple sessions to ensure reliability and to interpret barbell acceleration data at various intensities. Therefore, the primary purpose of this study was to assess the test – retest reliability of a portable accelerometer system in measuring the peak barbell acceleration of national-level weightlifters over three training sessions. The secondary purpose was to investigate the effects of different training intensities on peak barbell acceleration. It was hypothesized that the potable accelerometer system would show high test –retest reliability in barbell

526 K. Sato et al. acceleration over multiple training sessions and the peak barbell acceleration would show a gradual decline as the overall intensity increases from 80% to 90% of 1RM.


Methods Men’s and women’s weightlifting resident team members at the Colorado Springs Olympic Training Center participated in this study (N ¼ 12). The athletes’ demographic profiles are shown in Table I. The athletes were free of injuries at the time of data collection. The data were collected during the period of their strength development phase leading up to a competition. This study was conducted in compliance with policies of the United States Olympic Committee on the testing of athletic participants. A triaxial accelerometer (PS-2119, Pasco Scientific, Roseville, CA, USA) was used to measure barbell acceleration, and was attached to a Bluetoothe wireless device (Pasco Pasport Airlink SI (PS-2005)). Recently, published data reported that this device measured acceleration as accurately as a high-speed camera at the same sampling rate (Sato et al., 2009b). The total mass of the unit is 170.1 g. In order to minimize external shock when lifters drop the barbell, a foam unit was designed to secure the accelerometer to the end of the bar (Figure 1). The total mass of the unit plus the foam was 240.3 g, which is equivalent to a metal barbell collar. The accelerometer unit was placed on the right side of the bar, relative to a lifter, and the metal collar was placed on the left side of the bar during data collection. Thus, the weight of the accelerometer should not interfere with a lifter’s ability to sense asymmetry of weight between the left and right sides of the barbell. The foam protection was necessary due to the nature of the sport; lifters drop the bar after completion of a lift. It is also important to note that the orientation of the sensor had to remain in a constant position relative to the bar throughout the lift to avoid aberrant signal from the resultant acceleration. The unit was placed directly underneath the barbell prior to each trial (Figure 1). All participants reported to the training facility of USA weightlifting for data collection, and were provided the testing procedure. The athletes performed a self-selected bout of stretching and warm-up as they normally do before a training session. A specific warmup included static and dynamic stretching of the whole body, back squats, and weightlifting movements with light weights leading up to 80% of 1RM. The testing intensity of 80%, 85%, and 90% of each athlete’s 1RM was based on their best records in the snatch at previous official meets. The identical procedure was used to collect these data over three different days (seven days apart). During the pilot test, the investigator could be up to 30 m away from the lifter and capture the acceleration data wirelessly without loss of signal. For this study, the investigators were positioned approximately 5 m away from the platform. The barbell acceleration data were collected by performing a snatch lift for three repetitions at an intensity of 80%, 85%, and 90% of 1RM on three different test dates.

Table I. Descriptive data for participant characteristics (M^ SD).

Age (year) Body height (m) Body mass (kg) Training experience (year)

Male (n ¼ 7)

Female (n ¼ 5)

23.0 ^ 3.4 1.78 ^ 0.11 98.0 ^ 24.0 7.3 ^ 1.7

20.0 ^ 1.4 1.63 ^ 0.10 71.8 ^ 16.7 5.8 ^ 1.1



527

Figure 1. Accelerometer attachment to the barbell.

Data were collected with DataStudioe software (Version 1.9, Pasco Scientific) which acquires, displays, and analyzes the data. A sampling rate of the accelerometer was set at 100 Hz. Recent studies showed that 100 Hz is an appropriate sampling rate to capture weightlifting motions (Sato et al., 2009a, 2009b). The acceleration data are displayed in ‘real-time’ for immediate feedback. This tri-axial accelerometer summed x, y, and z-axes of acceleration for data analysis. The data automatically removed gravity of 9.8 m/s2, as the stationary or constant velocity displays zero acceleration. A typical trend of barbell acceleration of the snatch is shown in Figure 2. A previous pilot study showed that the highest barbell acceleration is typically generated during the second pull phase of the snatch among experienced weightlifters and collegiate athletes. The second pull is also a critical phase of weightlifting where forces are exerted to lift the bar over-head for the snatch (Garhammer & Hatfield, 1985; Stone et al., 2006a). Based on this information, our focus of the barbell acceleration measurement has been the instantaneous point of the second pull, where the highest barbell acceleration is achieved. To address the first research question, each intensity level of the data was averaged using three repetitions for each lifter (Henry, 1967). Then the values were analyzed additionally across three testing days for intra-class correlation coefficient (ICC) to obtain test –retest reliability (Kroll, 1967; Carmines & Zeller, 1979; Hopkins, 2000). To address the second research question, each intensity level was averaged using all lifters’ data and analyzed with a one-way repeated measure ANOVA to compare the peak barbell acceleration on various intensities ( p ¼ 0.05). The SPSSw Predictive Analytics SoftWare was used for the analyses (SPSS version 17: An IBM company, New York, NY, USA).


528 K. Sato et al.

Figure 2. Barbell acceleration data of snatch.

Results Test – retest ICC of the peak barbell acceleration results over the multiple training sessions was r ¼ 0.88, with 95% confidence interval of 0.81 –0.93. Similar data outcome between the different training sessions was shown with high ICCs of r ¼ 0.95 between training day 1 and 2, r ¼ 0.86 between training day 1 and 3, and 0.85 between training day 2 and 3. A one-way repeated measure ANOVA was calculated by comparing the intensity levels of 80%, 85%, and 90% of the 1RM. Statistical significance was found (F2,10 ¼ 25.34, p , 0.01) with an effect size of 0.84 from h 2 calculation. Peak barbell acceleration was 19.96 ^ 2.69 m/s2 at 80%, 15.98 ^ 2.73 m/s2 at 85%, and 13.73 ^ 3.15 m/s2 at 90% of 1RM with coefficient of variance in 13.5%, 17.1%, and 23.0%, respectively. Followup paired sample t-tests revealed that the peak barbell acceleration decreased significantly from 80% to 85% ( p , 0.01), from 80% to 90% ( p , 0.01), and from 85% to 90% ( p , 0.01). Discussion The primary purpose of the study was to track peak barbell acceleration from three different intensities over three training sessions to obtain ICCs for test –retest reliability. A previous study (Sato et al., 2009b) noted that the acceleration device was valid based on high correlations of barbell acceleration data with a high-speed camera at the same sampling rate (100 Hz), but it was not tested for test – retest reliability over several different days of data collection. The results of this study addressed the concern by showing relatively high ICCs (95% confidence interval of 0.81 –0.93), which indicated that each lifter’s peak barbell acceleration at three intensities was consistent on multiple training sessions. Within the 15-day period (day 1, day 8, and day 15 of data collection), it is possible that lifters can get



529

stronger from the daily training over 15 days. Based on this study, peak barbell acceleration appeared to remain stable with the same load during the 15-day window. And it is assumed that strength gain over the 15 days was minimal and therefore did not affect the peak barbell acceleration. However, it is important to note that the small deviation of variability within a lifter over multiple trials may exist due to lifting mechanics variation. Overall, the high ICC values in this study support the notion that the device is reliable over multiple days of data collection. This device may be useful for barbell acceleration tracking purposes over short or long periods of training. The potential findings may become extremely important for monitoring athletes over a long period. The secondary purpose was to compare different training intensities to examine how they influence peak barbell acceleration. The results of the study supported the hypothesis that barbell acceleration decreased as the mass of the bar increased from 80% to 85% and from 85% to 90% of 1RM. It is logical to conclude that peak barbell acceleration decreased due to increasing load. An interesting observation captured during the pilot study noted peak barbell acceleration stayed relatively constant from 50% to 80% of 1RM during our initial pilot study. This is one reason to collect the present data from 80% of 1RM. This consistency indicates that the changes in barbell mass from 50% to 80% of 1RM did not affect the peak barbell acceleration, but increasing mass did change the pattern of acceleration when intensity exceeded 80% of 1RM. It appears that approximately 80 –85% of 1RM could be the threshold for decreasing acceleration of the barbell for these lifters. Furthermore, the acceleration decline could also be interpreted that the lifter’s force production applied to the barbell reaches near maximal. In this study, the peak force data were not reported. The peak force values among lifters varied due to extreme difference in amount of weights they tested. This was caused by the weight class difference (i.e. 53 kg female vs. 105þ kg male lifter), then the peak force at each intensity greatly varied. For this reason, future studies need to identify appropriate statistical procedures to address this issue. Moreover, future studies need to test participants who are relatively in the same weight class to minimize variation in the peak force. Further investigation is essential to understand how strength increase influences the peak acceleration over long-term monitoring on athletes. Expectation would be an increase in peak acceleration with strength gain. Increased force production may lead to maintained peak acceleration at higher loads, or the peak acceleration may be higher. Previous studies explained the importance of measuring peak power (Garhammer & Hatfield, 1985; Casatelli et al., 2010; Kraemer, 2010). Like other portable devices, this unit does not measure power output; however, acceleration is perhaps the single unique variable that, if measured easily, could serve as a vital feedback tool in sport tasks. The investigators were interested at the instantaneous point of peak barbell acceleration during the bar path at the second pull phase of the snatch. Reporting averaged acceleration value was another option but it may omit important information about a key period of acceleration. Although there is no literature supporting that the second pull is more important than the other phases of the lift, the instantaneous point of the second pull appears more important when assessing the explosiveness to lift the bar for catch. Previous studies displayed a graph of only vertical acceleration in the snatch (Garhammer & Hatfield, 1985; Gourgoulis et al., 2000). This study collected and resolved data as the resultant acceleration (x, y, and z). Investigators in this study believe that it is important to capture resultant acceleration because a typical bar movement approximates an S-shaped curve in the sagittal plane (Isaka et al., 1996; Gourgoulis et al., 2000; Haff et al., 2003; Gourgoulis et al., 2004). Especially, during the second pull, it is less likely to capture bar trajectory as purely vertical. Therefore, resultant acceleration may be more accurate measure


530 K. Sato et al. than just vertical acceleration. However, a limitation needs to be mentioned about a lack of knowledge in difference between single-axis versus tri-axes acceleration measures. The present data were taken from nationally high-ranked weightlifters. The Olympic training center is one of few locations to collect such data for research purpose. As the accelerometer is valid and reliable, taking the device to elsewhere to collect data from larger samples in future. Olympic weightlifting movements and variations (power snatch/power clean) are well utilized in the strength and conditioning field. The benefits of accelerometry for coaches evaluating lifts are many. Previous study showed the importance of immediate visual and verbal feedback to improve power snatch technique (Winchester et al., 2009). Accelerometer data could be used in a similar way. Additionally, attempting lifts of maximal weight are one way to measure how athletes improve their strength over long term, but measuring the peak barbell acceleration can be another useful assessment to observe progression of an athlete’s force production capability. Another benefit is that when tracking the peak barbell acceleration throughout a single training session, significant decreases in acceleration for later stages of a training session could be an indicator of fatigue (i.e. less force is being applied to accelerate the barbell). In weightlifting training, and arguably in any sport, over-reaching or fatigue is a critical measure of adaptation. However, there is no scientific evidence on a practical, non-invasive way to measure it. Force is another way to measure performance and it is proportional to acceleration, the accelerometer may allow us the ability to indirectly measure fatigue in a practical way in the training environment. If the athlete continues to lift after becoming substantially fatigued, this may lead to poor training adaptation and over-training/overuse injuries. Describing and identifying fatigue are difficult, but barbell acceleration may be a suitable assessment for detecting fatigue via declining acceleration values. This idea is consistent with one study, reporting that barbell velocity of the clean decreased without an adequate amount of rest (Haff et al., 2003). Therefore, an appropriate period of rest between repetitions and sets is a key to delay fatigue and perhaps enhance performance. Conclusion The accelerometer was reliable in tracking peak barbell acceleration over multiple sessions. With the validation of the previous study, the unit accurately measured a critical component of weightlifting, and it is possibly useful for performance assessment. Future research may attempt to obtain normative data on barbell acceleration such that lifters can assess themselves against a ‘gold standard.’ If this concept holds true, it would be appropriate to compare the data and gauge progress among lifters of all abilities. Although further investigations are merited, this study provides some insight into the benefits that barbell accelerometry measurement may have for weightlifting performance.

References Baechle, T. R., & Earle, R. W. (2000). Essentials of strength training and conditioning: Resistance training. Champaign, IL: Human Kinetics. Barton, J. (1997). Are there general rules in snatch kinematics? Paper presented at the Weightlifting Symposium, Ancient Olympia, Greece. Bartonietz, K. E. (1996). Biomechanics of the Snatch: Toward a higher training efficiency. Strength and Conditioning Journal, 18 (3), 24–31. Carmines, E. G., & Zeller, R. A. (1979). Reliability and validity assessment. Newbury, CA: Sage.



531

Casatelli, N., Muller, R., & Maffiuletti, N. A. (2010). Validity and reliability of myosystem for the assessment of vertical jump height. Journal of Strength and Conditioning Research, 24 (11), 3186–3193. Garhammer, J., & Hatfield, F. G. (1985). Weightlifting. In T. K. Cureton (Ed.), Encychopedia of physical education, fitness, and sport. Reston, VA: AAHPERD. Gourgoulis, V., Aggeloussis, N., Mavromatis, G., & Garas, A. (2000). Three-dimentional kinematic analysis of the snatch of elite Greek weightlifters. Journal of Sports Sciences, 18, 643–652. Gourgoulis, V., Aggeloussis, N., Kalivas, V., Antoniou, P., & Mavromatis, G. (2004). Snatch lift kinematics and bar energetics in male adolescent and adult weightlifters. The Journal of Sports Medicine and Physical Fitness, 44, 126– 131. Haff, G. G., Whiteley, A., McCoy, L. B., O’Bryant, H. S., Kilgore, J. L., Haff, E. E., Pierce, K., & Stone, M. H. (2003). Effects of different set configurations on barbell velocity and displacement during a clean pull. Journal of Strength and Conditioning Research, 17 (1), 95–103. Henry, F. M. (1967). “Best” versus “Average” individual scores. The Research Quarterly, 38, 317–320. Hiskia, G. (1997). Biomechanical analysis on performance of world and Olympic champion weight lifters. Paper presented at International Weightlifting Federation Coaching, Referring & Medical Symposium, Ancient Olympia, Greece. Hopkins, W. G. (2000). Measures of reliability in sports medicine and science. Sports Medicine, 30, 1–15. Isaka, T., Okada, J., & Funato, K. (1996). Kinematics analysis on the barbell during the snatch movement of elite Asian weight lifters. Journal of Applied Biomechanics, 12, 508–516. Kraemer, J. W. (2010). Construct validity of the Myotest (R) in measuring force and power production. Journal of Strength and Conditioning Research, 24, 1. Kroll, W. (1967). Reliability theory and research decision in selection of a criterion score. The Research Quarterly, 38, 412– 419. Lee, Y. H., Huwang, C. Y., & Tsuang, Y. H. (1996). Biomechanical characteristics of preactivation and pulling phases on snatch lift. Journal of Applied Biomechanics, 11, 288–298. Sato, K., Fleschler, P., & Sands, W. A. (2009a). Barbell acceleration analysis on various intensities of weightlifting. Paper presented at International Society of Biomechanics in Sports Conference, Limerick, Ireland. Sato, K., Smith, S. L., & Sands, W. A. (2009b). Validation of accelerometer for measuring sport performance. Journal of Strength and Conditioning Research, 23 (1), 341–347. Schilling, B. K., Stone, M. H., O’Bryant, H. S., Fry, A. C., Coglianese, R. H., & Pierce, K. (2002). Snatch technique of collegiate national level weightlifters. Journal of Strength and Conditioning Research, 16 (4), 551– 554. Stone, M. H., O’Bryant, H. S., Williams, F. E., Pierce, K., & Johnson, R. L. (1998). Analysis of bar paths during the snatch in elite male weightlifters. Strength and Conditioning Journal, 20 (4), 30–38. Stone, M. H., Pierce, K., Sands, W. A., & Stone, M. E. (2006a). Weightlifting: A brief overview. Strength and Conditioning Journal, 28 (1), 50– 66. Stone, M. H., Pierce, K., Sands, W. A., & Stone, M. E. (2006b). Weightlifting: Program design. Strength and Conditioning Journal, 28 (2), 10– 17. Winchester, J. B., Porter, J. M., & McBride, J. M. (2009). Changes in bar path kinematics and kinetics through use of summary feedback in power snatch training. Journal of Strength and Conditioning Research, 23 (2), 444– 454.