Abstract
BACKGROUND/OBJECTIVES: The food frequency questionnaire (FFQ) is a commonly used tool for assessing the usual diet of groups and individuals. This study developed a chatbot-based mobile FFQ, embedded within KakaoTalk, Korea's most popular mobile instant messenger. This study compared the energy and nutrient intake reported by participants using a chatbot versus a paper-based FFQ. SUBJECTS/METHODS: This study included 95 participants (50 men and 45 women, aged 24-79 yrs) who underwent gastroscopy or colonoscopy for cancer screening at Dongguk University Ilsan Hospital between April and August 2022. The participants completed both the chatbot and paper-based FFQs within a 3-mon period. The comparability of the chatbot-based FFQ with traditional paper-based FFQ was examined by comparing the log-transformed intakes of energy and nutrients using Pearson correlation coefficients. The Cohen's Kappa coefficients, Bland-Altman plots, and cross-classification percent agreement were used to evaluate the compatibility. RESULTS: The Pearson correlation coefficients of energy and energy-adjusted nutrients ranged from 0.74 (niacin) to 0.90 (vitamin A), with a median coefficient of 0.85. The Cohen kappa coefficients varied from 0.42 (niacin) to 0.64 (n-6 fatty acid). The Bland-Altman plots suggested that most of the data points fell between the lower and upper limits of agreement, and 88% to 98% of the participants were classified into the same or adjacent quartiles for energy-adjusted nutrients. CONCLUSION: These findings indicate good comparability between chatbot- and paper-based FFQs. A chatbot-based mobile FFQ can provide comparable dietary intake rankings, making it a viable tool for evaluating the associations between diet and various disease risks in longitudinal studies.