Related Experiment Video
Updated: Jun 18, 2026
Clinical Anthropometrics and Body Composition from 3-Dimensional Optical Imaging
Published on: June 7, 2024
Accuracy, repeatability, and interchangeability of smartphone-based digital anthropometry using multi- and dual-image
Irismar Gonçalves Almeida da Encarnação1, Matheus Santos Cerqueira2, Grant M Tinsley3
1Department of Physical Education, Federal University of Viçosa, Viçosa, Minas Gerais, Brasil; Academic Department of Education, Federal Institute of Southeast of Minas Gerais, Campus Rio Pomba, Rio Pomba, Minas Gerais, Brazil.
Background & Aims:
Digital anthropometry using mobile applications (DAMs) has emerged as an accessible alternative for estimating body composition. However, its accuracy and repeatability remain uncertain, and it is unclear whether different DAMs provide comparable results or whether image acquisition approach (dual- and multi-image approaches) influences performance. This study aimed to evaluate the repeatability and accuracy of DAMs for estimating body fat percentage (BF%) against dual-energy X-ray absorptiometry (DXA), as well as to assess their interchangeability and compare the performance of dual- and multi-image approaches.
Methods:
Fifty-seven healthy adults (30 men, 27 women; 28.0 ± 10.2 years) underwent BF% estimations using three DAMs (MeThreeSixty, Bodygram, and ZOZO Fit) and DXA (reference) under standardized conditions. DAMs were selected based on free availability, novelty, and image-capture approach, including a multi-image DAM (ZOZO Fit, twelve photographs per scan) and two dual-image systems. Test-retest repeatability was assessed via intraclass correlation coefficient (ICC) and coefficient of variation (CV%). Accuracy was evaluated through mean differences, concordance correlation coefficient (CCC), root mean square error (RMSE), and Bland-Altman analysis. Interchangeability among DAMs was also examined.
Results:
All DAMs showed excellent test-retest repeatability (ICC = 0.98-0.99; CV ≤ 4.1%). When compared with DXA, all DAMs significantly underestimated BF%, with mean biases ranging from -1.2% for ZOZO Fit to -2.1% for Bodygram. ZOZO Fit, the only multi-image system, demonstrated the lowest estimation error (RMSE = 2.95%), highest concordance with DXA (CCC = 0.90). However, none of the DAMs achieved statistical equivalence within ±2% of DXA, and all showed wide limits of agreement in all comparisons.
Conclusions:
DAMs exhibited excellent repeatability but low accuracy and were not interchangeable. Although the multi-image DAM showed relatively superior performance at group level analysis, none of the evaluated DAMs currently supports use for population-level monitoring, clinical diagnosis, or individual-level decision-making. Further validation, calibration, and correction for systematic bias are required before broader practical use.
