Challenges and Application of IngID, a Framework for Parsing and Systematic Reporting of Ingredients Used in Commercially Packaged Foods. (7th June 2021)

• Record Type: Journal Article
• Authors: Nguyen, Quynhanh; Li, Ying; Ahuja, Jaspreet; Pehrsson, Pamela
• Is Part Of: Current developments in nutrition. Volume 5(2021)Supplement 2
• Page Start: 600
• Page End: 600
• Publication Date: 2021-06-07
• Subjects: Nutrition -- Periodicals; Nutritional Physiological Phenomena; Nutrition; Periodicals; Periodicals; Fulltext; Internet Resources; Periodicals; 612.3
• DOI: 10.1093/cdn/nzab044_031
• Languages: English
• ISSNs: 2475-2991

Abstract

Abstract: Objectives: The purpose is to describe the IngID's challenges and application of the development process. Methods: Ingredient lists were mainly obtained from BFPD for top sellers of each food category. Manufacturer's and third-party websites were used if the ingredients from BFPD were not available. Ingredient lists were then compiled and parsed using a custom program in R, a software application. The program identified unmatched parentheses where numbers of open and closed parentheses were not matched. Another step involved in parsing ingredients from ingredient lists on the labels was manually developing replacement files for ingredient clarification. After parsing and assigning preferred descriptors (PDs), the PDs were then assigned Broader Term (BT) based on types or functions such as sweeteners, pertinent to specific research questions. Results: Challenges during the process included missing parentheses and complicated ingredient lists especially in Frozen and Refrigerated Entrees and Snack bars categories. There were ∼710 ingredient lists in these 3 categories with unmatched parentheses. Assigning BTs could be difficult as one PD can be assigned different BTs; potato starch could be assigned as a starch, stabilizer, or thickener. An example of IngID's application was assigning BTs for types of sweeteners. There were 101 different sweeteners which were assigned to 4 different BTs: nutritive and non-nutritive sweeteners, reduced-calorie sweeteners, and fruit-juice concentrates. Conclusions: The IngID program is still evolving, especially with respect to automation of some of the pre-processing steps to save time and avoid human errors. Once fully developed, IngID will help identify and report commercial ingredients, and will be useful for developing artificial intelligence tools, research, and food ontology. Funding Sources: Agricultural Research Service, U.S. Department of Agriculture.