summary: Researchers have developed a machine learning algorithm, FoodProX, that can predict the degree of processing in food.
This tool scores foods on a scale of 0 (minimally or unprocessed) to 100 (highly ultra-processed). FoodProX fills a gap in existing nutrient databases and provides high-resolution analysis of processed foods.
The development is an important advance for researchers investigating the health effects of processed foods.
Important facts:
- FoodProX is a machine learning tool that predicts food processing levels.
- This tool utilizes nutritional information from the USDA Food and Nutrient Database.
- AI tools have confirmed that over 73% of the U.S. food system is ultra-processed.
sauce: Northeastern University
Researchers in the Northeast are keen to better understand the relationship between “ultra-processed foods” and human health through the University-sponsored Foodome project.
As part of that effort, researchers at the Center for Complex Networks Research have now developed a machine-learning algorithm that accurately predicts the degree of processing of the foods that make up the U.S. food supply.
their discovery Nature Communications in April.
A machine learning classifier, called FoodProX, uses nutrition label information provided by the USDA Dietary Research Foods and Nutrients Database as input and scores the level of processing of specific foods.
The algorithm works by producing an output representing the likelihood that each food item falls into one of four categories that are part of the NOVA food classification system (a system developed by researchers at the University of São Paulo, Brazil). Works. “Widely used in epidemiological studies”.
Users can try out the tool by visiting the TrueFood research project website. Users can search for foods and see their food processing scores. The algorithm assigns each product a single score ranging from 0 (indicating “minimal or unprocessed” food) to 100 (highly ultra-processed food).
Using FoodProX, researchers were able to fill a gap in the nutrient database for dietary research. Categorize “complex recipes and mixed foods and meals.” We offer high resolution lenses for inspecting processed foods.
As a result, FoodProX will provide a clearer understanding of what processed foods are really like, an important step for researchers studying the health effects of these foods, the researchers said. doing.
Researchers believe that the NOVA system, which classifies foods into four categories ranging from “unprocessed or minimally processed” to ultra-processed, is fundamentally He points out that there are limits.
“This perception of homogeneity in NOVA 4 foods limits both scientific research and practical consumer guidance on the health effects of different degrees of processing,” the researchers wrote.
“It will also reduce industry incentives to reformulate food towards less processed products, shifting investment from ultra-processed NOVA 4 foods to less processed NOVA 1 and NOVA 3 categories.”
“What we’re really saying in this paper is that we believe that nutritional information, the chemicals that are measured as nutrients in nutrient profiles, somehow encode traces of food processing.” said Julia Menichetti, a senior research fellow at Northeastern University Network Sciences. She is the research institute and lead author of the study.
“Because when we process food, when we change some of the main ingredients, its chemistry changes in different ways.”
Its ‘fingerprinting’ is a way that researchers can glean insight into how much chemical changes have been made to a particular food.
“We don’t necessarily know what all the chemical fingerprints associated one-to-one with each process are,” Menichetti told Northeastern Global News. “I can’t even count how many ways food can be processed.”
Ultimately, the AI tool corroborated the team’s previous finding that more than 73% of the U.S. food system is ultra-processed, providing a previously unavailable level of detail. Menichetti says her team has succeeded for the first time in creating her AI tool that reliably assesses the chemical content of food.
“This is the first paper in the fields of nutrition and public health to leverage machine learning to reproducibly and systematically score foods according to their degree of food processing,” she says.
The team’s work is important because, as Menichetti puts it, “there hasn’t been much of a data culture” in the nutrition and health sciences related to food processing, providing a scientific perspective on what processing means. This was because it encouraged a less rigorous discussion.
“Without a systematic way to observe foods and assess their properties, it’s difficult to conduct comparable large-scale studies elsewhere in the world,” says Menichetti.
“FPro helps assess an individual’s dietary quality and provides predictive power for more than 200 health variables,” said Robert Gray Dodge, Professor of Network Sciences, Northeastern University, and author of the study. Co-author Albert Laszlo Barabasi said:
“This demonstrates the impact of replacing processed foods with less processed alternatives of the same item to create personalized dietary changes with minimal effort.”
About this machine learning research news
author: Tanner Stenning
sauce: Northeastern University
contact: Tanner Stening – Northeastern University
image: Image credited to Neuroscience News
Original research: open access.
“Machine learning prediction of food processing degree” Giulia Menichetti et al. Nature Communications
overview
Machine learning prediction of food processing degree
Despite accumulating evidence that increased consumption of ultra-processed foods has adverse health effects, it remains difficult to determine what constitutes processed foods.
Indeed, current processing-based food classifications are limited in scope and do not distinguish between degrees of processing, hindering consumer choice and delaying research on the health effects of processed foods.
Here we present a machine learning algorithm that accurately predicts the degree of processing of any food. This indicates that over 73% of his US food supply is ultra-processed.
We have found that increased reliance on ultra-processed foods in an individual’s diet correlates with metabolic syndrome, diabetes, angina, increased blood pressure, increased biological age, and decreased vitamin bioavailability. showed a decline.
Finally, we found that replacing foods with less-processed alternatives can significantly reduce the health impacts of ultra-processed foods, and that access to information about how processed foods are not currently available to consumers could help the public. It suggests that it may improve your health.
