‘True food’ research database offers rankings for 50,000 processed foods

A study by investigators from Mass General Brigham provides information to empower consumers and policymakers about the degree of processing of the foods available at three large grocery retailers.

Using an algorithm, the researchers analyzed the offerings from these stores and assigned a score to foods based on how processed they were. They also analyzed food categories to determine what options are available at different stores.

Their results are published in Nature Food.

“There are a lot of mixed messages about what a person should eat. Our work aims to create a sort of translator to help people look at food information in a more digestible way,” said corresponding author Giulia Menichetti, Ph.D., an investigator in the Channing Division of Network Medicine at Brigham and Women’s Hospital.

“By creating a system of scoring processed food, consumers don’t have to be overwhelmed with excessive and challenging information to be able to eat healthier.”

In the last few decades, highly and “ultra” processed foods have flooded supermarket shelves. These foods often contain additives, preservatives and high amounts of sugar, salt, and unhealthy fats.

Research has suggested that diets high in ultra-processed foods can contribute to health issues like obesity, diabetes and heart disease. Over-processing can also strip foods of beneficial nutrients. But there is no easy way to tell what foods are processed, highly processed, or ultra-processed. Researchers also know very little about what foods are available in real-life scenarios.

“When people hear about the dangers of ultra-processed foods, they ask, ‘OK, what are the rules? How can we apply this knowledge?'” Menichetti said.

“We are building tools to help people implement changes to their diet based on information currently available about food processing. Given the challenging task of transforming eating behaviors, we want to nudge them to eat something that is within what they currently want but a less-processed option.”

The researchers gathered and analyzed the ingredient lists, nutrition facts and prices of foods available in three leading U.S. grocery stores: Target, Whole Foods, and Walmart. They used this data to create GroceryDB, a database with over 50,000 food items.

They published the results on the TrueFood website, which consumers can use to compare foods. The website features a processing score, nutrition facts, and an ingredient tree that shows the makeup of various foods.

The researchers then used their FPro algorithm, published in a 2023 article in Nature Communications, which leverages machine learning to determine the degree of processing for the available foods and captured in a “processing score.”

The higher the score, the more ultra-processed a food is. While Whole Foods offers more minimally processed options, most of the food these stores sell is ultra-processed.

In some stores, highly processed foods were the only option in some categories. For example, cereals at Whole Foods cover a range of FPro values—from minimally- to ultra-processed. However, all cereals available at Walmart and Target had a high processing score. This same trend was seen in the soups and stews, yogurt and yogurt drinks, milk and milk substitutes, and cookies and biscuits categories.

The authors note that while grocery stores may sell a large variety in terms of quantity of products and brands, the offered processing choices can be identical in multiple stores, limiting consumer nutritional choices to a narrow range.

While the data in GroceryDB and on the True Food website is remarkably detailed, it remains limited because it originates from just three stores at a single point in time. In the future, the researchers would like to add geolocation information and temporal monitoring to learn about food options in different areas of the country and how that variability may impact the social determinants of health.

“People can use this information, but our goal would be to push this to become a large-scale, data-driven tool to improve public health,” Menichetti said.

“Most research activities in nutrition still depend on manual curation, but our study shows that artificial intelligence and data science can be used to scale up.

“This initiative not only lays the groundwork for similar efforts globally, but also underscores the critical role of open-access, internationally comparable data in advancing global nutrition security with the goal of ensuring that everyone has access to health food options in their daily activities.”

A study by investigators from Mass General Brigham provides information to empower consumers and policymakers about the degree of processing of the foods available at three large grocery retailers.

Using an algorithm, the researchers analyzed the offerings from these stores and assigned a score to foods based on how processed they were. They also analyzed food categories to determine what options are available at different stores.

Their results are published in Nature Food.

“There are a lot of mixed messages about what a person should eat. Our work aims to create a sort of translator to help people look at food information in a more digestible way,” said corresponding author Giulia Menichetti, Ph.D., an investigator in the Channing Division of Network Medicine at Brigham and Women’s Hospital.

“By creating a system of scoring processed food, consumers don’t have to be overwhelmed with excessive and challenging information to be able to eat healthier.”

In the last few decades, highly and “ultra” processed foods have flooded supermarket shelves. These foods often contain additives, preservatives and high amounts of sugar, salt, and unhealthy fats.

Research has suggested that diets high in ultra-processed foods can contribute to health issues like obesity, diabetes and heart disease. Over-processing can also strip foods of beneficial nutrients. But there is no easy way to tell what foods are processed, highly processed, or ultra-processed. Researchers also know very little about what foods are available in real-life scenarios.

“When people hear about the dangers of ultra-processed foods, they ask, ‘OK, what are the rules? How can we apply this knowledge?'” Menichetti said.

“We are building tools to help people implement changes to their diet based on information currently available about food processing. Given the challenging task of transforming eating behaviors, we want to nudge them to eat something that is within what they currently want but a less-processed option.”

The researchers gathered and analyzed the ingredient lists, nutrition facts and prices of foods available in three leading U.S. grocery stores: Target, Whole Foods, and Walmart. They used this data to create GroceryDB, a database with over 50,000 food items.

They published the results on the TrueFood website, which consumers can use to compare foods. The website features a processing score, nutrition facts, and an ingredient tree that shows the makeup of various foods.

The researchers then used their FPro algorithm, published in a 2023 article in Nature Communications, which leverages machine learning to determine the degree of processing for the available foods and captured in a “processing score.”

The higher the score, the more ultra-processed a food is. While Whole Foods offers more minimally processed options, most of the food these stores sell is ultra-processed.

In some stores, highly processed foods were the only option in some categories. For example, cereals at Whole Foods cover a range of FPro values—from minimally- to ultra-processed. However, all cereals available at Walmart and Target had a high processing score. This same trend was seen in the soups and stews, yogurt and yogurt drinks, milk and milk substitutes, and cookies and biscuits categories.

The authors note that while grocery stores may sell a large variety in terms of quantity of products and brands, the offered processing choices can be identical in multiple stores, limiting consumer nutritional choices to a narrow range.

While the data in GroceryDB and on the True Food website is remarkably detailed, it remains limited because it originates from just three stores at a single point in time. In the future, the researchers would like to add geolocation information and temporal monitoring to learn about food options in different areas of the country and how that variability may impact the social determinants of health.

“People can use this information, but our goal would be to push this to become a large-scale, data-driven tool to improve public health,” Menichetti said.

“Most research activities in nutrition still depend on manual curation, but our study shows that artificial intelligence and data science can be used to scale up.

“This initiative not only lays the groundwork for similar efforts globally, but also underscores the critical role of open-access, internationally comparable data in advancing global nutrition security with the goal of ensuring that everyone has access to health food options in their daily activities.”