Validation of Artificial Intelligence-based Digital Anthropometry Applications for Estimating Body Composition.

Overview

Anthropometric measurements are commonly employed to evaluate body composition, morphology, and health-related parameters across diverse populations. While a cost-effective and field-friendly method, the COVID-19 pandemic has spurred research on digital anthropometry worldwide. Machine learning, a fusion of artificial intelligence and data mining, holds promise for enhancing data collection and analysis in Kinanthropometry applications. Rather than replacing traditional methods, digital anthropometry presents a significant opportunity to enhance accuracy, validity, practicality, and the implementation of self-monitoring procedures under professional guidance. The CyberMetron Project by DBSS aims to perform additional research and increase scientific literacy among practitioners for public awareness of digital anthropometry.

Anthropometric measurements are commonly used to assess body composition, morphology, and health-related parameters in diverse populations. Although it is an economical and practical method in the field, the COVID-19 pandemic has propelled worldwide research on digital anthropometry. Machine learning, a fusion of artificial intelligence and data mining, promises to enhance the collection and analysis of data in applications that rely on cineanthropometric data. Instead of replacing traditional methods, digital anthropometry represents a significant opportunity to improve the precision, validity, practicality, and implementation of self-monitoring procedures under professional supervision. The CyberMetron Project by DBSS aims to conduct additional research and increase scientific literacy among professionals to raise awareness about digital anthropometry. Data will be collected from residents of both genders and with different levels of physical activity. The population sample will be obtained through internal calls to both students and administrative staff at participating universities and through an open call. Anthropometric variables of restricted and complete profiles established by ISAK will be measured, along with digital images taken at different distances from the lens. Regarding the sample size, it will be conducted by convenience (non-probabilistic), primarily considering university students, administrative staff, and other potentially eligible adults who respond to the study announcement and sign the informed consent. Finally, all statistical analyses will be performed within the statistical computing environment R v4.2.3, with a statistical significance of P\<0.05. Pearson's correlation coefficient (r), adjusted determination coefficient (aR²), and Lin's concordance correlation coefficient (ρc) will be used for comparative analysis between main variables taken by conventional anthropometry and digital anthropometry. The coefficient of repeated measures correlation (rrm) will be employed to assess the strength of the linear association between variables, while the intraclass correlation coefficient (ICC), with its corresponding 95% confidence interval (95% CI), and Finn's coefficient (rF) will be used to evaluate reliability between evaluators. Additionally, the Bland-Altman analysis will be applied for the concordance analysis between conventional and digital anthropometry.