Dr. Marie G Gantz, PhD

Personalized Nutrition

for Health Optimization

The goal of this Nutrition for Precision Health (NPH) powered by All of Us research study is to develop Artificial Intelligence/Machine Learning (AI/ML) algorithms that predict individual responses to diet patterns using rich multimodal data streams collected across multiple domains (e.g., behavior, social, environmental, clinical and molecular biomarkers). NPH includes a large phenotyping cohort (Module 1, N=8000) and two separate follow-up groups drawn from a subset of Module 1participants. One group (Module 2, N=1200) receives three distinct diets in a 14-day crossover sequence, with at least a 14-day washout period between diets, while living in their own homes. A second group (Module 3, N=150) receives the same three diets under full-time supervision in a residential research setting. We will train and test AI/ML models to predict 0-4 hour postprandial response curves for glucose, insulin, triglycerides, and GLP-1, to the standardized diet-specific meal test (DSMT) collected after each of the three different diets delivered in Module 2. Each diet functions as a controlled stimulus to reveal biological features (such as individual variables, patterns, or clusters of measurements) that best predict a person's response. The Module 2 DSMT response curves are the primary outcomes (dependent variables) for AI/ML algorithms that predict individual responses to diet patterns. As a secondary objective, NPH will evaluate the validity and acceptability of technology-based dietary assessment tools. The Automated Self-Administered 24-hour recall (ASA24), Automatic Ingestion Monitor-2 (AIM-2), and the mobile food record (mFR) will be evaluated in Modules 2 and 3, and the ASA24 food record and the image-assisted ASA24 recall will be evaluated only in Module 3. Total energy intake, macronutrient and dietary fiber intake data are the main outcomes for validity testing compared against measures of actual intake. Acceptability will be determined from feedback surveys.