(OR15-05-24) Development of a Method to Identify the MyPlate Food Groups for Plant-Based Alternative Products

Objectives: Recent analysis of plant-based meat and milk alternative products has revealed high variability in their nutrient content. We wondered if this variability would affect how we should classify these plant-based alternatives into existing food group categories, such as those used by MyPlate. Methods are needed for plant-based alternatives to be accurately incorporated into food-group-based dietary guidance. We aimed to use a data-driven approach to compare plant-based alternatives with MyPlate food groups.

Methods: Using food-level data from the University of Minnesota Nutrition Coordinating Center’s (NCC) Food and Nutrient Database, we identified two sets of foods. First, foods from MyPlate food groups and some subgroups that are not intended as alternatives to animal products. Food groups included were dairy, protein foods, vegetables, fruit, and grains. Second, plant-based alternatives intended to mimic animal products, including categories of meat, milk, yogurt, cheese, and ice cream and other frozen desserts. Using nutrient data converted to percent daily values per serving for 30 nutrients, we conducted dimensionality reduction using principal component analysis and tested for differences between food groups with permutational analysis of variance (PERMANOVA). Food groups were clustered for visualization using t-distributed stochastic neighbor embedding (t-SNE). Clusters were qualitatively assessed to identify if plant-based alternatives cluster with the foods they aim to mimic.

Results: MyPlate food groups were nutritionally distinct (PERMANOVA, p < 0.05, 99 permutations). Plant-based alternative products were similarly distinct by category (PERMANOVA, p < 0.05, 99 permutations). When plant-based alternatives were compared directly with MyPlate food groups, most of the plant-based milks formed a defined cluster between dairy cheese and dairy milk. Many plant-based cheeses clustered near dairy cheese and the grain and protein food groups. Most plant-based meats clustered within the grain foods and near the animal protein foods, while other plant-based meats clustered tightly with beans, peas, and lentils and near nuts and seeds.

Conclusions: A data-driven approach has the potential be useful when aiming to define the appropriate food group category for plant-based alternative products.

Funding Sources: None