(P02-085-25) Potential of Automatic Ingestion Monitor (AIM-2) for Dietary Assessment in Rural and Urban Children in Ghana

Objectives: We assessed agreement in measuring portion size, energy, macronutrients, fiber, and iron from the same meals using AIM-2 (camera worn on eyeglasses), image-assisted 24-hour recall (IA24-HR), and researcher-weighed food records (rWFR). Potentially, AIM-2 could address the limitations of traditional dietary assessment methods. We hypothesized that there would be good agreement between AIM-2 and the other methods in rural and urban children.

Methods: This cross-sectional study used purposive sample of 29 children under five years (CU5) (12 rural, 17 urban). AIM-2, IA24-HR, and rWFR were used to measure dietary intake on 2 weekdays and 1 weekend day. Nutritional analysis was conducted using custom AIM Annotation Software, programmed with West African, FNDDS, Kenyan, Ugandan, and USDA food composition databases. Seven response variables were examined by meals in which data were available for all three methods: consumed weight (g), energy (kcal), carbohydrate (g), fat (g), protein (g), fiber (g), and iron (mg). Agreement was assessed by calculating concordance correlation coefficients (CCC) and performing a Bland-Altman analysis.

Results: Mean (SD) ages (years) were 2.1 (1.0) for rural CU5 and 3.6 (0.7) for urban CU5. Overall, agreement was significant between all method-pairs and the ranges of CCC measured by location were: AIM-2 vs rWFR: rural (0.44 to 0.67), urban (0.22 to 0.54), and IA24-HR vs rWFR: rural (0.41 to 0.76), urban (0.25 to 0.74). Bland Altman’s plots showed good agreement between methods, although AIM-2 and IA24-HR consistently underestimated intakes. The lowest mean differences from the plots were found mainly in macronutrients: AIM-2 vs rWFR: carbohydrate (rural -0.25, urban -0.14), fat (rural -0.38, urban -0.10), protein (rural -0.41, urban -0.20); IA24-HR vs rWFR: carbohydrate (rural -0.29, urban -0.14), fat (rural -0.15, urban 0.12), protein (rural -0.18, urban 0.11).

Conclusions: Overall, there was some level of agreement between the methods. Agreement was notably good between AIM-2 versus rWFR in rural CU5, indicating the potential of AIM-2 for objectively assessing dietary intake in Ghanaian CU5.

Funding Sources: Bill & Melinda Gates Foundation and Agricultural Experimental Station