Standardized Data Systems Could Advance AI-Driven Food Research

A product developer attempting to create a plant-based protein drink for three international markets often faces a fragmented information landscape. One document may identify an ingredient as “haldi,” while another labels it “turmeric” and a third uses the scientific name “Curcuma longa.” Because these specifications are locked in separate PDFs, spreadsheets, and emails, developers spend significant time verifying data rather than focusing on product formulation.

The Growing Complexity of Food Safety

Managing this data volume is increasingly critical as food development becomes more complex. According to 2026 estimates from the World Health Organization regarding 42 foodborne hazards, unsafe food caused an estimated 866 million illnesses and 1.5 million deaths globally in 2021. Chemical hazards were responsible for 73% of these fatalities, with inorganic arsenic and lead accounting for 42% and 31% of chemical-related deaths, respectively. The WHO also reported that productivity losses from foodborne diseases reached $310 billion in 2021.

The challenge intensifies when companies operate in multiple regions. Each market enforces distinct allergen regulations, contaminant limits, and labeling requirements. Ingredients, particularly botanicals used in wellness products, vary based on their source, harvest season, and processing methods. As companies add new suppliers and ingredients, the volume of information requiring verification expands.

Implementing Shared Vocabularies

Industry experts suggest that centralized databases are not the sole solution. Instead, organizations are turning to ontologies—systems that provide a shared vocabulary with built-in relationships. FoodOn, an open food ontology described in npj Science of Food in 2018, categorizes thousands of food products, sources, and processing methods. By using such systems, companies can enable artificial intelligence to recognize that different terms refer to the same ingredient, linking it automatically to its test results, supplier lots, and market requirements.