Validation of nutrient intake estimated by a smartphone-based dietary recording application among older adults

Validation studies of smartphone-based dietary recording applications among older adults remain limited. This study evaluated the validity of nutrient intake estimates derived from the meal-check feature (MCF) of the online Kayoinoba application, a smartphone-based community health promotion application for older adults. Thirty-one community-dwelling older adults aged ≥ 65 years in Japan (mean age 70.4 years) recorded their usual diet for 10 days using the MCF and completed handwritten, weighed dietary records (WDR) on 5 of those 10 days. Five-day average intakes were compared using Bland–Altman analysis, two one-sided tests for equivalence (± 10% margin), mean differences, Pearson’s correlations, intraclass correlation coefficients (ICCs), and cross-classification agreement. Equivalence was confirmed for energy, protein, fat, carbohydrate, folate, magnesium, iron, and zinc. Percentage differences were perfect or sufficient for almost all nutrients (mean percentage 0.1–9.1%). Correlations and ICCs were moderate to strong for most nutrients, whereas salt equivalent showed weaker individual-level agreement. Cross-classification showed high agreement (77.4–100%). The MCF showed favorable validity for estimating several key nutrients and may provide a practical, low-burden option for dietary monitoring in older adults.