Body composition assessment is a relevant element in the biomedical field, in research and daily practice in the medical and nutritional fields, and in the management of athletes. This paper aimed to operate in an Italian sample investigating the possibility of predicting the somatotype from bioimpedance analysis and comparing the predicted results with those obtained from anthropometric measurements. This observational study was conducted with retrospective data collected from 2827 subjects. The somatotype of each subject was calculated both with the Heath−Carter method and by a multiple regression model based on bioimpedance and anthropometric parameters. Somatotypes (endomorph, mesomorph, and ectomorph) were predicted with a high goodness of fit (R2 adjusted > 0.80). Two different somatocharts were obtained from anthropometric measures and bioimpedance parameters and subsequentially compared. Bland−Altman plots showed acceptable accuracy. This study could be a first step in developing a new approach that allows the detection of a subject’s somatotype via bioimpedance analysis, stratified according to sex, with a time-saving and more standardized procedure. It would allow, for example, during the COVID-19 pandemic, to minimize operator−patient contact in having measurements.
Keywords: bioimpedance; body composition; linear models; somatotypes.