Artificial Intelligence, Modeling, and Informatics for Nutrition Guidance and Systems (AIMINGS) Center - Abstract – Overall AIMINGS Center
The vision of this proposed Artificial Intelligence, Modeling, and Informatics, for Nutrition Guidance and
Systems (AIMINGS) Center is to implement computational and data science approaches and tools to advance
nutrition for precision health in a way that accounts for the complex systems involved. Many existing data sets
include extraneous data, making them difficult to analyze at best, and at worst, prone to generating misleading
or biased insights. Thus, there is a need to for new approaches, methods, and tools to collapse and distill data
to make them more Artificial Intelligence (AI)-ready and ready for a range of different analyses. This coincides
with the goal of Project 1: to develop and utilize The Data Distiller for Precision Nutrition, a set of
approached and tools that can collapse and distill nutrition-relevant data to create datasets that are AI-
ready and ready for a range of other analyses. The first objective of the Nutrition for Precision Health (NPH)
program is to “examine individual differences observed in response to different diets by studying the
interactions between diet, genes, proteins, microbiome, metabolism and other individual contextual factors.”
Given the type of missing data we face in nutrition, and the importance of establishing causal relationships
rather than correlations, there is a need for new imputation methods. To address this, Project 2, the Causal
Relationship Disentangler, will introduce new approaches for handling missing data while preserving
causal structure. Learning how to transfer causal knowledge and doing so with missing data is critical
for realizing the potential of nutrition for precision health. The NPH program’s other objectives are “to use
AI to develop algorithms to predict individual responses to foods and dietary patterns,” and “to validate
algorithms for clinical application.” This requires bringing different causal pathways together to understand how
they interact. Agent-based models (ABMs) can help and serve as "virtual laboratories" to predict how different
people may respond to a particular diet under different circumstances. Therefore, the goal of Project 3 (The
Virtual Human for Precision Nutrition) is to develop an ABM tool that can help better understand and
predict an individual's response to food and dietary patterns, while bringing together and accounting
for the interactions between genetic, physiological, and behavioral factors. However, focusing on the
individual alone will not be enough to address all aspects of NPH. Therefore, the Virtual Public Health
Precision Nutrition Laboratory (Project 4) will develop ABMs that represent and account for the
systems outside individuals such as their social, economic, and built environments. An Administrative
and Coordination Core will oversee all operations and a pilot program. A Data Systems Core (DSC) will
leverage the substantial computing resources of CUNY, West Point, and the Department of Defense to create
a flexible cloud-based architecture for data flow and a collaborative workspace. A Computational Systems
Core will provide resources and personnel to support the DSC and tool development/deployment.
