ABSTRACT
The prevalence of overweight and obesity has gradually increased throughout the world. In this sense, body composition evaluation' methods have gained special attention in some research areas. The aim of this study was to propose a mathematical model to estimate body fat percentage (BF%) of men from thermal imaging and body circumferences variables and using as gold standard values obtained by Dual-Energy XRay Absorptiometry (DXA). This was a cross-sectional study. A Sample of 47 men aged from 18 to 20 years was involved in the study. They were selected among soldiers from Brazilian Army. Data collection was carried out in just one day, in the following sequence: thermal imaging, body fat percentage assessed by DXA and anthropometric measures. It was performed the identification of the skin temperature strongest correlated with BF% in each body segment (trunk, upper and lower limbs). The same procedure was done with anthropometrics variables. After that, these variables were tested in the linear regression model. Waist, abdomen and hip circumferences, as expected, showed strong positive correlations with BF% (r = 0.826, 0.873, and 0.853, respectively). Since the skin temperature values tended to negative correlation with BF%, the best results were found for mean skin temperature of the anterior face of the trunk (r = -0 753) and posterior face (r = -0723). The best model reaches an R square of 0.800 using just one anthropometric and one skin temperature variable (BF% =32.34 + (0.6 x AC) - (1.87 x ATSkTMean)), where AC means abdominal circumference and ATSkTMean means the mean skin temperature from the anterior face of the trunk. The proposed model can be used by Brazilian Army to estimate BF% of soldiers and male military personal aged from 18 to 20 years.
