Machine Learning Prediction of Food Processing
Abstract
In recent years the classification of the food supply has become essential to public health dietary guidelines, assisting the population in adopting a healthy diet. NOVA, a classification focusing on the extent of food processing, has enabled many epidemiological studies investigating the association between ultra-processed food consumption and disease onset, despite the strong dependence on manual assessment.
In this talk, I will introduce FoodProX, a machine learning algorithm able to predict NOVA labels for food databases with varying nutrient resolution. FoodProX allowed us to analyze all food items, as well as the consumption data, across several cycles of nationally representative datasets provided by the National Health and Nutrition Examination Survey from 1999 to 2018. Our analysis of several food databases showed how discrete classes only partially captured the processing heterogeneity of the food supply, prompting us to define the ``food processing score'', a continuous index that ranks all foods from the least processed to the most ultra-processed, and defining degrees of processing within each NOVA category. We show that this score can also be extended to measure the overall quality of individuals’ diet, including its relation to major diet-related health parameters.
Speaker Bio
Giulia is a physicist, with a background in network modeling of biological information. She currently leads the Foodome project, which aims to track the full chemical complexity of the food we consume and develop quantitative tools to unveil, at the mechanistic level, the impact of these chemicals on our health.