Examining the validity of Diet ID™ for measuring Healthy Eating Index

BackgroundDiet ID™ is a digital, image-based dietary assessment tool that uses pattern recognition and comparisons of meal images to estimate overall dietary intake and diet quality.ObjectiveTo examine the relative validity of Diet ID™ in measuring Healthy Eating Index-2020 (HEI-2020) total and component scores.MethodsIn this cross-sectional study, participants completed Diet ID™ and the Diet History Questionnaire II, a semiquantitative food frequency questionnaire. Spearman's rho correlations examined relative validity. Wilcoxon signed-rank tests, cross-classification, and Bland-Altman plots assessed agreement.ResultsThe HEI-2020 total scores were moderately correlated (r_s = 0.46, p < 0.001) and most components were weakly correlated (r_s = 0.12-0.34, p < 0.05). There was poor agreement between the dietary assessment methods, as evidenced by significant differences in most scores (p < 0.001), fewer than half of the participants being classified into the same tertile, and Bland-Altman plots.ConclusionFurther work is needed to improve Diet ID™ to measure HEI-2020 total and component scores.