(P06-004-25) WISEcode Ultra-Processed Food (Wc-UPF™): A Data-Driven Precision Tool for Evaluating and Categorizing Processed Foods

Objectives: To develop the WISEcode Ultra-Processed Food (Wc-UPF™) system as a data-driven and actionable framework for assessing processed foods, addressing limitations of broad categorization systems like NOVA, particularly its ultraprocessed food  classification.

Methods: The Wc-UPF system was developed iteratively across four models, each adding to the previous. Model 1 used a simple count of processed ingredients. Model 2 weighted each ingredient score based on processing degree and possible health concern. Model 3 included % total energy from added sugars. Model 4 included a short list of ingredients with known  safety concerns. Foods were scored and categorized into five levels: minimal, light, moderate, ultra, and super-ultra. The scoring system was applied to and validated against a database of 100,243 commercial foods, covering 138 different grocery product categories.

Results: Model development revealed strengths and limitations at each stage. Model 1 failed to distinguish between simple foods with multiple ingredients vs heavily processed foods with few ingredients. Model 2 improved differentiation, but did not capture high sugar load foods or beverages despite the presence of additives. Model 3 accounted for the sugar load but failed to capture highly concerning additives. Model 4 successfully differentiated among all processed foods. The resulting distribution of commercial foods by Wc-UPF™ level was relatively balanced: 19.9% minimal, 16.4% light, 22.0% moderate, 18.9% ultra, and 22.8% super-ultra. The model resulted in a level of granularity absent in NOVA classifications while ensuring that foods with credible concerns, such as those containing banned additives, are classified appropriately regardless of numeric score.

Conclusions: Wc-UPF™ provides a nuanced, ingredient-informed tool for categorizing processed foods, by assessing degree of ingredient processing, potential health impacts, and added sugar content. This approach can support dietary recommendations, product reformulation, and informed consumer choices, while its modular design allows for continuous refinement by incorporating new research findings and predictive factors.

Funding Sources: This work was funded by WISEcode, LLC.