Morphology, body composition and maturity status of young Colombian athletes from the Urabá subregion: A k-Medoids and hierarchical clustering analysis

Abstract

The Uraba subregion is one of the most prominent cradles of Colombian elite athletes and, therefore, highly recognized within the “Land of Athletes” framework of the Colombian Ministry of Sports. In order to contribute to the young talent identification and selection of sports specialization, the aim of this STROBE-based cross-sectional study was to determine the morphological characteristics (MC), body composition (BC) and maturity status (MS) of U16 athletes from this subregion (7 municipalities). Eighty-one young athletes (66 weightlifters, 15 boxers) with at least one regional-competition of experience participated (33F; 48M; 14.9 ± 1.4 years; 62.28 ± 16.6 kg; 162.8 ± 9.9 cm). After parental informed consent, ISAK-standardized anthropometric data were collected during a youth sports championship. Athletes were subdivided in clusters using the PAM (k-Medoids clustering) and the bottom-up agglomerative (hierarchical clustering) algorithms. Both clustering methods were based on 55 variables that encompassed MC (raw variables, indices, somatotype), BC (five-compartment model, %BF-equations, ΣS) and MS (maturity offset, PHV, inter alia). The number of clusters was predefined as k = 2 since was the best solution according to 18 criterion-algorithms (100 bootstrap simulations). Non-parametric tests showed significant differences between sex, sports, municipalities and clusters for certain analysed variables. Internal validity of the clustering showed that sport type might explain the variation in the data; thus, it is noteworthy reasonable to recommend the implementation of unsupervised machine learning strategies along with other supervised methodologies in the identification and characterization of young talents and early sports specialization in Colombian athletes with Olympic projection but further research and support is needed.