S.3.49 - Artificial Intelligence as Applied Predictive Models for Food Composition Databases
Saturday, June 17, 2023 |
8:15 AM - 9:30 AM |
Clarion Hotel Gillet - Room Akademien |
Details
Rationale:
Development and evaluation of nutrition public health policy interventions toward mitigating non-communicable diseases require a comprehensive database of foods. However, existing national food composition databases (FCD) are expensive to develop and challenging to maintain. Traditionally, FCD have relied on time-consuming and arduous method of food identification, categorization and estimation of nutritional quality to understand the food supply. Artificial Intelligence applied predictive modelling have the potential to improve the performance of labor-intensive tasks and to automate many of the manual processes in understanding the nutritional quality of the food supply. This symposium presents the current and future applications of Artificial Intelligence in food composition databases, from food categorization and identification to nutritional quality estimation.
Objectives:
1) To illustrate the use of artificial intelligence methods applied to automating tasks for food composition databases.
2) To explore the current and future applications of nutrition-focused AI models in monitoring and informing interventions for a dynamic food supply
3) To assess the applied predictive modelling and compare different AI methods
4) To demonstrate the potential of AI in addressing nutrition-related public health policies aimed at creating supportive food environments.
Summary:
This symposium will bring together results from Canada, Australia and Slovenia in developing and applying AI-predictive modelling to illustrating their widespread use in monitoring and informing interventions for the food supply. Through this symposium, we hope to facilitate a thought-provoking discussion on the potential of AI methods in evaluation of nutrition interventions in a timely manner by reducing the time and increasing the speed in labour-intensive tasks applied to food composition databases.
Format:
Introduction: Mavra Ahmed (5 minutes)
First Presenter: Gordana Ispirova, Slovenia (15-18 minutes)
Second Presenter: Guanlan Hu, Canada (15-18 minutes)
Third Presenter: Tazman Davies, Australia (15-18 minutes)
Discussion: Mary L'Abbe (15 minutes)
Interaction (if online):
Interaction will be facilitated in an online symposium but asking participants to 1) either input questions using the chat function or 2) have presenters prepare discussion questions in advance.
Speaker
Chair
Discussant
Predictive Modeling for Nutrient Prediction combining data and domain-driven knowledge
Machine learning approaches for food categorization and nutrient profiling
An innovative machine learning approach to predict the dietary fiber content of packaged foods
