Dietary Assessment by Pattern Recognition: a Comparative Analysis

Abstract

BackgroundDiet quality photo navigation (DQPN) is a novel dietary intake assessment tool that was developed to help address limitations of traditional tools and to easily integrate into health care delivery systems. Prevailing practice is to validate new tools against approaches that are in wide use. ObjectiveThis study aimed to assess 1) the validity of Diet ID in measuring diet quality, food group and nutrient intake against 2 traditional dietary assessment methods (i.e., food record [FR], food frequency questionnaire) and 2) the test reproducibility/reliability of Diet ID to obtain similar results with repeat assessments. MethodsUsing a participant-sourcing platform for online research, we recruited 90 participants, 58 of whom completed DQPN, a 3-d FR (via the Automated Self-Administered 24-hour Dietary Assessment Tool), and a food frequency questionnaire (FFQ, via the Dietary History Questionnaire III). We estimated mean nutrient and food group intake with all 3 instruments and generated Pearson correlations between them. ResultsMean age (SD) of participants was 38 (11) y, and more than half were male (64%). The strongest correlations for DQPN when compared with the other 2 instruments were for diet quality, as measured by the Healthy Eating Index 2015; between DQPN and the FFQ, the correlation was 0.58 (P < 0.001), and between DQPN and the FR, the correlation was 0.56 (P < 0.001). Selected nutrients and food groups also showed moderate strength correlations. Test-retest reproducibility for measuring diet quality was evaluated for DQPN and showed a correlation of 0.70 (P < 0.0001). ConclusionsThe current study offers evidence that DQPN is comparable to traditional dietary assessment tools for estimating overall diet quality. This performance, plus DQPN’s ease-of-use and scalability, may recommend it in efforts to make dietary assessment a universal part of clinical care.