PhD A.A. Pomerantsev¹
PhD V.A. Kashkarov^1
1Lipetsk State Pedagogical University named after P.P. Semenov-Tyan-Shansky, Lipetsk

Keywords: movement biomechanics, sport techniques, blunders in motor skills, sprint, associative rules

Background. Modern sport science gives a high priority to every success factor analysis, with the sport-specific motor skills being commonly perceived as critical for success in most of the modern sport disciplines [6].  Requirements to the technical fitness test and correction toolkits have notably grown lately. Visual assessments, considered the most fast and efficient, give the means to track and correct only serious technical errors detectable by the non-instrumental observations [2, 3]. Quality analyses in this case are clearly subjective i.e. determined by the individual perceptions and the sporting/ coaching experiences; whilst the modern cutting-edge biomechanical test systems still take too long and, hence, inefficient for the training process improvement purposes [7].

Body movements need to be considered and analyzed in every component on a systemic basis, with minor changes in every component triggering revisions of the technical execution on the whole [1]. We believe that the associative rules provide a sound framework to track correlations of the movement parameters and analyze the sport techniques on a systemic basis [8] to accurately and comprehensively assess the technical performance and find the best training tools to perfect the motor skills.

Objective of the study was to analyze benefits of a new sport technique analyzing toolkit that combines the visual technical performance scores with the technique rating data processed by a mathematical statistics toolkit. The technical performance analyzing tool was expected to be at least as accurate as the instrumental movement biomechanics tests in terms of the test data quality and as fast as the visual assessments by experts.

Methods and structure of the study. The study included expert valuations of the sprint techniques by 10 leading active coaches. Sampled for the study were the 17-18 year old academic sprinters (n=10) from the first-year course of the Physical Education and Sports Department of Lipetsk State Pedagogical University named after P.P. Semenov-Tyan-Shansky. The sample was tested by a maximal-speed sprint test with a 25m startup section, with only two strides fixed by FastecInLineс videocamera (250 frames per sec, 640x480 pixels) [5]. The 10 leading active coaches (including 2 Honored and 2 Top Class Coaches) were expected to score the sprint techniques on the whole and fix the following 16 typical errors: (1) low thigh; (2) passive propulsion of the pelvis; (3) short push-up; (4) wrong foot placement; (5) heal-to-tip stepping; (5) excessive tension; (7) wide elbows; (8) wrong transverse arm action; (9) short backward arm action; (10) too deep lean forward; (11) backward trunk deviation; (12) bent-leg ground contact; (13) low chin; (14) strained hands; (15) slow thigh crossing action; and (16) too high chin.

The expert findings were fixed in the study form with indication of the above errors to produce a total technical score. The experts were offered the video replays in a sequential manner, with due provisions to prevent mutual influences, revisions and corrections of the scores; and they were given no background information on the technique kinematics and dynamics. The expert valuations were summarized in an input data array of 10х10х17=1700 inputs. For the further analysis, each entry was formatted as follows: athlete and expert names (text); points (number); and errors (with the correlation logics). The data array was processed by a standard mathematical statistics toolkit to find the associative rules.

Results and discussion. A prior analysis of the data array found poor correlations of the expert findings, with the Kendal concordance value of W=0.29 – indicative of the wide variations in the coach’s opinions: see Table 1 hereunder. Knowing that the visual scores are clearly subjective i.e. determined by individual perceptions and sporting and coaching experiences, we assumed that the subjectivity may be leveled down to a degree by the summations and mathematical processing. Each of the 16 errors was also assumed to have a different effect on the sprint time albeit the correlation analysis found no significant differences in their contributions to the total scores.

Table 1. Expert valuations of the sprint techniques

Note: * expert score on a 10-point scale/ rank

When analyzing the latent correlations of the biomechanical parameters, we tried to find the relevant associative rules i.e. regularities in the errors. With this purpose the input data array was converted into a matrix of events (errors) and transactions (athletes’ names) with 583 errors detected in 10 athletes by 10 experts) – using Deductor Studio Academic (www.basegroup.ru) tools.

Given on Figure 1 hereunder is the matrix of associative rules with the causes and effects, with every rule described by the following logical sequence: Y follows from X: when X is found, it will be followed by Y. This means in our case that when some technical error is fixed, it is followed by some other error with the specified degree of probability. Such correlations found by analysis are commonly referred to as the associative rules.

It is natural that such correlated errors are not always fixed together, and we introduced the support rate and reliability rate to find correlations. In this case, the support rate in the associative rule is the number of athletes tested with both of the errors at a time. The reliability rate is indicative of the likelihood of the second error when the first one is fixed.

Having analyzed the data array to find simple combinations (of one technical error associated with the other), we found 238 associative rules with 42 of them verified by 100% support rates and 100% reliability rates. Errors with the high support and reliability rates are clearly detectable by experts and easily explainable by the modern sports biomechanics. Associative rules with the lowest support and reliability rates, as well as those totally uncorrelated, are of particular interest for analysts since it means that such errors are incompatible i.e. mutually exclusive.

Figure 1. Excerpt from the output matrix of associative rules found from the two-dimensional data array

It is not improbable that the technical performance analysis may find a combination of a few errors (4 at most) correlated with some other combination. Despite some complexity of such analyses, their products may be interesting in many aspects. Benefits of the found associative rules may be demonstrated by a step-by-step consideration of the specific technical errors. Thus above Figure 1 gives the associative rules with the 100% support rate for error #6. We found 51 associative rules for this particular error, of the 252 maximal-support and maximal-reliability rules (of the 3,508 rules in total). It is obvious that all these rules cannot be described and explained in detail herein, although we believe that the approach as such may be highly beneficial for application in training systems for movement biomechanics analyses and technical progress [4].

Conclusion. The study demonstrates benefits of the new technical performance analyzing method based on expert valuations verifiable by the associative rules and movement biomechanics as it gives the means to put the movement biomechanics analysis on a sound qualitative basis. Correlations of the technical errors found by the new method give the means to improve the training system efficiency by the timely and focused revisions.

References

1. Kvashuk P.V., Semayeva G.N., Maslova I.N. Faktorny analiz biomekhanicheskikh pokazateley grebli muzhchin-baydarochnikov [Factor analysis of biomechanical rowing indicators of male kayakers]. Uchenye zapiski universiteta im. P.F. Lesgafta. 2015. no. 9 pp. 114-116.
2. Korenberg V.B. Osnovy kachestvennogo biomekhanicheskogo analiza [Basics of high-quality biomechanical analysis]. M.: Fizkultura i sport, 1979. 208 p.
3. Korshikov V.M., Pomerantsev A.A. Biomekhanika [Biomechanics]. Study guide. Lipetsk: LSPU publ., 2012. 240 p.
4. Ovchinnikov Yu.D., Minchenko V.G. Biomekhanika dvigatelnoy deyatelnosti: metodika obucheniya studentov [Biomechanics of motor activity: student teaching methodology]. Fizicheskaya kultura. Sport. Turizm. Dvigatelnaya rekreatsiya. 2017. vol. 2. no. 4. pp. 47–51.
5. Pomerantsev A.A. Issledovaniya po sportivnoy biomekhanike s primeneniem optiko-elektronnykh metodov registratsii parametrov dvizheniya [Research on sports biomechanics using optical-electronic motion parameters recording methods]. Lipetsk: LSPU publ., 2018. 233 p.
6. Samsonova A.V. Vozmozhnosti primeneniya faktornogo analiza dlya resheniya zadach v oblasti fizicheskoy kultury i sporta [Application potential of factor analysis to solve problems in physical education and sports]. Proc. final res.-pract. conf. of SPbSUPE satff for 2005. St. Petersburg, 2006. pp. 21-22.
7. Suchilin N.G., Arkaev L.Ya., Savelyev V.S. Pedagogiko-biomekhanicheskiy analiz tekhniki sportivnykh dvizheniy na osnove programmno-apparatnogo videokompleksa [Educational and biomechanical analysis of sports movement techniques based on software and hardware video complex]. Teoriya i praktika fiz. kultury. 1996. no. 4. pp. 12-20.
8. Furaev A.N. Obosnovanie znacheniy biomekhanicheskikh parametrov ryvka shtangi s pomoshchyu assotsiativnykh pravil [Justifying values ​​of biomechanical parameters of clean and jerk using association rules]. Uchenye zapiski universiteta im. P.F. Lesgafta, 2015. no. 12 (130). 277-281.

Corresponding author: ldclipetsk@mail.ru

Abstract

Modern sport science gives a high priority to every success factor analysis, with the sport-specific motor skills being commonly perceived as critical for success in most of the modern sport disciplines. Objective of the study was to analyze benefits of a new sport technique analyzing tool that combines the visual quality analysis with the technique rating data processed by a mathematical statistics toolkit. The study included expert valuations of the sprint running techniques by 10 leading active coaches who rated the skills of 10 elite sprinters. The experts were expected to fix typical errors if any and score the sprint techniques on the whole in 17 aspects. The expert scores were verified by objective mathematical statistic tools to find contradictions in the expert scores and ratings. Since every expert used 17 variables to rate the sprint techniques of the 10 athletes, the data array totaled 1700 values processed by the mathematical statistics toolkit. The analytical method tested by the study had made it possible to find 3508 associative rules to rate the modern sprint techniques including 252 core rules with the maximal meaning and dependability rates. The study demonstrates benefits of the new motor skill analyzing method based on expert valuations verifiable by the associative rules and movement biomechanics.