The American journal of clinical nutrition · 2024
Computer vision-assisted dietary assessment through mobile phones in female youth in urban Ghana: validity against weighed records and comparison with 24-h recalls.
СтатьяСравнительное исследованиеВалидационное исследованиеПоддержка исследования: вне правительства СШАНазвание и аннотация: Перевод в очереди
Обработка статьи · этапы, ошибки и управление
Загрузка состояния…
Ошибки и ограничения
История · было → стало
Состояние обновляется каждые 5 секунд, пока блок открыт. Процент показывается только там, где известен объём работы. Пауза останавливает продолжение этапа; текущий запрос к источнику может завершиться.
Gelli Aulo, Nwabuikwu Odiche, Bannerman Boateng, Ador Gabriel, Atadze Vicentia, Asante Millicent, Bempong Silas, McCloskey Peter, Nguyen Phuong Hong, Hughes David, Folson Gloria
Аннотация
BACKGROUND: Gaps persist in the data on diets and on the validity of dietary assessment methods in youth in low-income and middle-income countries (LMICs) due to costs constraints. Although computer vision-assisted dietary assessment tools have been proposed, limited evidence exists on their validity in LMICs.
OBJECTIVES: This study aimed to validate FRANI (Food Recognition Assistance and Nudging Insights), a mobile phone application with computer vision-assisted dietary assessment, against weighed records (WRs) and compare with 24-h recalls (24HR), in female youth in Ghana.
METHODS: Dietary intake was assessed on 2 nonconsecutive days using FRANI, WR, and 24HR in females aged 18-24 y recruited at the University of Ghana, Accra (n = 64). Equivalence was examined by comparing intake mean ratios (FRANI/WR and 24HR/WR) with error margins of 10%, 15%, and 20%, using mixed-effect regression models adjusting for repeated measures. Agreement between methods was assessed using the concordance correlation coefficient (CCC).
RESULTS: Equivalence for FRANI and WR was found at 10% bound for riboflavin and vitamin B-6 intakes and at 15% bound for protein, fat, calcium, folate, iron, thiamine, vitamin C, and zinc intakes. Energy, fiber, vitamin A, and niacin intakes were equivalent at 20% bound. Comparisons between 24HR and WR found no estimates within a 10% bound. Protein, iron, niacin, riboflavin, and zinc intakes were equivalent at a 15% bound; folate, thiamine, and vitamin B-12 intakes were equivalent at a 20% bound. CCCs between FRANI and WR ranged from 0.45 to 0.74 (mean: 0.60) and between 24HR and WR ranged from 0.48 to 0.76 (mean: 0.63). Omission errors were 15% for FRANI and 22% for 24HR. Intrusion errors were 22% for FRANI and 18% for 24HR.
CONCLUSIONS: FRANI-assisted dietary assessment accurately estimates nutrient intake and performed as accurately as 24HR in female youth in Ghana. Although improvements in computer vision-assisted diet assessment are possible, emerging evidence on FRANI suggests its readiness for scale-up.
OBJECTIVES: This study aimed to validate FRANI (Food Recognition Assistance and Nudging Insights), a mobile phone application with computer vision-assisted dietary assessment, against weighed records (WRs) and compare with 24-h recalls (24HR), in female youth in Ghana.
METHODS: Dietary intake was assessed on 2 nonconsecutive days using FRANI, WR, and 24HR in females aged 18-24 y recruited at the University of Ghana, Accra (n = 64). Equivalence was examined by comparing intake mean ratios (FRANI/WR and 24HR/WR) with error margins of 10%, 15%, and 20%, using mixed-effect regression models adjusting for repeated measures. Agreement between methods was assessed using the concordance correlation coefficient (CCC).
RESULTS: Equivalence for FRANI and WR was found at 10% bound for riboflavin and vitamin B-6 intakes and at 15% bound for protein, fat, calcium, folate, iron, thiamine, vitamin C, and zinc intakes. Energy, fiber, vitamin A, and niacin intakes were equivalent at 20% bound. Comparisons between 24HR and WR found no estimates within a 10% bound. Protein, iron, niacin, riboflavin, and zinc intakes were equivalent at a 15% bound; folate, thiamine, and vitamin B-12 intakes were equivalent at a 20% bound. CCCs between FRANI and WR ranged from 0.45 to 0.74 (mean: 0.60) and between 24HR and WR ranged from 0.48 to 0.76 (mean: 0.63). Omission errors were 15% for FRANI and 22% for 24HR. Intrusion errors were 22% for FRANI and 18% for 24HR.
CONCLUSIONS: FRANI-assisted dietary assessment accurately estimates nutrient intake and performed as accurately as 24HR in female youth in Ghana. Although improvements in computer vision-assisted diet assessment are possible, emerging evidence on FRANI suggests its readiness for scale-up.