Identificación de somatotipos con inteligencia artificial: medidas antropométricas vs bioimpedancia

Abstract

The shape of a person's body depends on their race due to genetic inheritance. The shape of a person's body is classified into somatotypes. The somatotypes are: endomorph, mesomorph and ectomorph. To determine the somatotype, the Heath and Carter method can be applied by measuring height, weight, arm and ankle diameters, as well as skin folds. The shape of the body is relevant in athletes. Therefore, in this work we try to predict the somatotype of a person. The prediction would be achieved without doing a calculation. A Dataset was obtained with values obtained from a bioimpedance device and anthropometric measurements. Algorithms were trained with both types of data using neural networks, SVM, and linear regression. Results were obtained and each scenario is compared. The best results were those generated with anthropometric measurements, reaching 96.8% accuracy