Unsupervised machine learning analysis of the anthropometric characteristics and maturity status of young Colombian athletes
Abstrakt
Abstract Introduction: The study of anthropometry-based indicators of morphology and maturity status might contribute to the talent identification, sports specialization and early categorization of young athletes. The Urabá subregion is considered one of the geographical locations with the highest sport potential in Colombia; however, no study has evaluated young combat athletes.
Objective: The aim of this study was to characterize for the first time morphology, body composition, and biological maturation status of pre-adolescent Olympic wrestling athletes from Urabá subregion (Antioquia, Colombia). Materials and methods: A STROBE-based cross-sectional study was carried out in forty-nine young Olympic wrestlers (20F: 29M; 13.3 +- 1.2 years; 154.0 +- 11.7 cm; 45.8 +- 10.8 kg; 19.0 +- 2.9 kg.m-2) with previous experience in sports events and competing in the Urabá regional games.
Anthropometry-based variables of morphology, body composition (five- and two-compartment models), and maturity status were analyzed. An unsupervised machine learning algorithm was used to identify similar data groups (clusters) and extract profile patterns.
Results: Several morphological, body composition and maturity status differences were found between girls and boys (p < 0.05). We identified two significantly different phenotypes representing lighter, shorter, leaner, more biomechanically efficient, and in late maturing (Cluster 1) versus taller, heavier, more robust, less biomechanically efficient, and average matured (Cluster 2) young athletes.
The matching analysis of the clusters revealed that maturity explained most of the variance in the data. Conclusions: Two clustering-based phenotypes were obtained to provide relevant information that might assist nutrition and exercise professionals when designing interventions.
More research is needed to evaluate potential associations with physical performance and/or sport success. Keywords: Anthropometry; Body Composition; Somatotypes; Biological maturation; Youth sports; Sports medicine; Cluster Analysis.