Anthropometric baseD Estimation of adiPoSity
The ADEPS project aims to estimate body fat based on a number of easily measured anthropometric body measures to provide a more accurate estimate of body fat compared with body mass index (BMI). The study uses body volume measurements that allow dividing the body according to the two-component model into fat and lean mass.
The first phase involved working with a dataset of existing 3D body scans of subjects. Based on anthropometric measures from these body scans, two calibration sets and a validation set were selected via a design of experiments (DOE) approach. A method was developed to calculate body volumes from the selected body scans and a protocol was established to eliminate scan noise at the level of the head, hands and feet. The second phase involved a measurement campaign in which adult subjects were measured with a 3D body scanner and an air displacement plethysmograph (BodPod®, gold standard for volume measurements). Additionally, an electrical resistance measurement was performed to determine hydration status and some anthropometric measures were measured manually.
The first phase involved working with a dataset of existing 3D body scans of subjects. Based on anthropometric measures from these body scans, two calibration sets and a validation set were selected via a design of experiments (DOE) approach. A method was developed to calculate body volumes from the selected body scans and a protocol was established to eliminate scan noise at the level of the head, hands and feet. The second phase involved a measurement campaign in which adult subjects were measured with a 3D body scanner and an air displacement plethysmograph (BodPod®, gold standard for volume measurements). Additionally, an electrical resistance measurement was performed to determine hydration status and some anthropometric measures were measured manually.
Resultaten
A total of 248 subjects (132 women) participated in the fieldwork of which 245 (129 women) were available for analysis. Two regression models were built for each sex with body fat percentage (LV%) and body fat (kg) obtained from the gold standard as dependent variables. The selection of the final models was done through an iterative process in which parameters of model quality and results of measurement system analyses supported the selection process. For LV%, the explained variance (R2) and estimation error (RMSE) in women and men are 77% (RMSE 4.34%) and 78% (RMSE 4.94%), respectively. In comparison, the explained variance of classical BMI for LV% is only 65% (RMSE 5.3%) in women and 58% (RMSE 6.4%) in men. For body fat (kg), the R2 values in women and men are 93% (RMSE 2.84 kg) and 88% (RMSE 3.87 kg), respectively.
The ADEPS project was publicised through various dissemination activities including the establishment of a project website (www.adeps.net), design of an icon cf media campaign HOGENT, web advertisement, vulgarising articles with wide international distribution, a scientific article and participation in and presentations at international conferences. The valorisation potential will be further investigated under the guidance of TechTransfer and in cooperation with the business developer of an IOF valorisation consortium.
The ADEPS project was publicised through various dissemination activities including the establishment of a project website (www.adeps.net), design of an icon cf media campaign HOGENT, web advertisement, vulgarising articles with wide international distribution, a scientific article and participation in and presentations at international conferences. The valorisation potential will be further investigated under the guidance of TechTransfer and in cooperation with the business developer of an IOF valorisation consortium.
01/10/2014 - 01/10/2016
