1

Introduction¹

Artificial intelligence (AI), machine learning (ML), and deep learning (DL) have shown promise toward aiding in developing algorithms to better understand and predict interactions between food- and nutrition-related data and health outcomes, particularly when large amounts of data need to be structured and integrated. However, additional research is needed to identify areas where AI/ML are likely to have an impact and their limitations. In addition, federal agencies are interested in exploring criteria around how to best use AI/ML in nutrition research.

To explore current knowledge and practice related to the application of advanced computation, big data analytics, and high-performance computing to support scientific advances in food and nutrition research, the National Academies of Sciences, Engineering, and Medicine’s (the National Academies’) Food and Nutrition Board convened experts to discuss this and related subjects in Washington, DC, on October 10–11, 2023. The speakers and participants discussed definitions and methods; the appropriate use of evidence generated from these methods to inform food- and nutrition-related programs and policies; considered issues related to diversity, equity, inclusion, bias, and privacy; identified opportunities and challenges related to capacity building and training; and explored the future potential of these

___________________

¹ The planning committee’s role was limited to planning the workshop, and the Proceedings of a Workshop was prepared by the workshop rapporteurs as a factual summary of what occurred at the workshop. Statements, recommendations, and opinions expressed are those of individual presenters and participants and are not necessarily endorsed or verified by the National Academies of Sciences, Engineering, and Medicine, and they should not be construed as reflecting any group consensus.

methods in food and nutrition research. The workshop sessions highlighted applications and lessons learned from studies of AI, ML, and DL methods in both food and nutrition research and other fields. Box 1-1 provides the statement of task for the workshop.

INTRODUCTORY REMARKS

Rodolphe Barrangou, the Todd R. Klaenhammer Distinguished Professor at North Carolina State University and workshop planning committee cochair, welcomed participants and said that the workshop would focus on the future of food and nutrition research and the role that advanced computation, predictive technologies, and big data analytics will play. “We have to talk about challenges and opportunities. We have to talk about building the systems we need to implement that technology,” said Barrangou.

Sharon Kirkpatrick, associate professor in the School of Public Health Sciences at the University of Waterloo and workshop planning committee

cochair, summarized what was ahead. The workshop would start by laying a solid foundation in terms of key concepts related to data science; introduce the theme of ethics, privacy, bias, and trust to be considered; explore how data science and AI/ML are being used in nutrition and food sciences; and outline some of the related promises and challenges. The first day would include a session on applications and lessons learned from work on wearables, the microbiome, and metabolomics. The day would end with a session on capacity building and inclusivity. Day 2 would include a second session on applications and lessons learned, focusing on designing nutrition studies for AI data analysis, how to gain farmers’ trust in AI, and the application of AI to supply chains. The following session would focus on the potential applications of AI and data science to large-scale initiatives. The final session would feature a broad discussion of the workshop’s key themes.

Institute for Advancing Health Through Agriculture

Patrick Stover, director of the Institute for Advancing Health through Agriculture (IHA) at Texas A&M University, said that IHA was created to use systems approaches to reimagine the connections between food and the health of the nation. IHA focuses on precision nutrition, understanding the variability in the diet and disease relationship, responsive agriculture, and healthy living. He defined responsive agriculture as “an agriculture system and food environment that supports health through nutrition for all while ensuring the system is economically robust and environmentally sustainable for future generations” and healthy living as “translating advancements in precision nutrition and responsive agriculture into evidence-based practices and policies to make food and agriculture the solution to skyrocketing health care costs.”

This is a critical time for nutrition science and public health nutrition, said Stover. Dietary patterns are a major driver of rising health care costs affecting everyone. Over 70 percent of people in the United States have overweight or obesity, and 60 percent have at least one chronic health condition. “But we also know that we can bring the very best science to bear to achieve solutions to agriculture, food, and nutrition,” he said, noting that agriculture has always responded to societal expectations. For example, agriculture and food systems were successfully engineered after World War II to produce calories in abundance, making hunger and food insecurity rare for most households and not the result of insufficient food production. In subsequent decades, research led to understanding nutritional deficiency disorders—“hidden hunger”—and developing population-based guidance and policies that largely prevented them.

The nation faces the challenge of addressing diet-related chronic diseases and has new expectations for food, agriculture, and nutrition. “Including health and chronic disease reduction as goals of food and agriculture will require transformational advances across the entire food and agriculture value chain,” said Stover. “The science and policies we use to address hunger and nutritional deficiency disorders are comparatively simple; the diet-related chronic disease connection includes multiple interacting health behaviors and environmental exposures.”

At the individual level, one size does not fit all regarding diet–disease or diet–health relationships, adding to the complexity of the challenge. Thus, the same approaches used to address nutritional deficiencies are inadequate to address the variability and dynamics that define the connection between agriculture, food, nutrition, and health. “Making food and agriculture the solution for chronic disease reduction will require new approaches, new types of data, and better ways of communicating dietary information of the public,” said Stover. “The expectations are high, and

the rigorous science we are going to address today must lead the way.” It is critical, he added, to avoid overpromising and get this right to maintain public support.

Stover said that although data science is transforming society and offering solutions to address complexity, the field of food and nutrition is a late adopter. However, advances in AI, including layering numerous associations on validated, physiological, metabolic, or social computational networks, offer the possibility of establishing true causal relationships that underpin connections among agriculture, food, and health.

U.S. Department of Agriculture (USDA) Agricultural Research Service (ARS) Human Nutrition Program

Cindy Davis, national program leader for the USDA-ARS Human Nutrition Program, said that the workshop’s topic is exceedingly relevant to her program, whose mission is to define the role of food and its components in optimizing health throughout the life cycle for all Americans by conducting high-national-priority research. AI/ML, she said, has shown promise for developing algorithms to better understand and predict interactions between food- and nutrition-related data and health outcomes, particularly when large amounts of data need to be structured and integrated. However, additional research is needed to identify areas where AI/ML is likely to have an impact and understand their limitations.

ARS, said Davis, is USDA’s chief in-house scientific agency focused on finding solutions to agricultural problems and conducting research on individual barriers to consuming a healthy diet and achieving a healthy body weight. Its six human nutrition research centers have a core capability for long-term, multidisciplinary, translational research in high-priority areas to improve the nation’s health. Its five priorities for 2024–2029 are

1. Bridging the gap between food production and human health by identifying the agricultural practices influencing the nutritional quality and composition of food and conducting multidisciplinary research to understand the complex interactions within the food system and their effects on human health.
2. Monitoring food composition and the nation’s nutrient intake to provide food composition data, determine national food consumption and dietary patterns, and develop improved methods to analyze food and determine food and nutrient intake.
3. Developing the scientific basis for dietary guidance by improving the scientific basis for updating national dietary standards and guidelines, identifying mechanisms whereby food, food components, and physical activity promote health, and using advanced

3. technology to develop and integrate multiple data sources to more precisely inform nutritional requirements.
4. Preventing diet-related chronic disease by identifying mechanisms by which food, food components, and physical activity can help and developing and evaluating diet and physical activity strategies.
5. Understanding life-stage nutrition and metabolism.

Davis noted an increasing recognition that understanding the connections and synergies between nutritional health and agriculture can be achieved only through the broad framework of food systems and simultaneous research across all pillars of the food system. Consensus is emerging that food systems contain four primary, interactive, and interdependent components: human nutrition/health, food production and agriculture, food technology’s effects on the environment and vice versa, and consumer choices and attitudes. Understanding the complex interactions within the food system related to human health requires multidisciplinary teams that assess inputs and effects from all sectors. Davis stated that advanced computation, predictive technologies, and big data analytics, of which AI and ML are examples, are necessary to integrate these data.

The U.S. food supply, said Davis, is fluid, and providing timely and accurate food composition data is complex because of constant changes in food regulations and policy, food choices and consumer preferences, production and processing methods that induce compositional variability, and demographic changes in the population. In addition, food composition and food intake data are only as accurate as the methods used to obtain them, making advances in instrumentation, analytical procedures, and methodology necessary to provide high-quality data.

Davis noted that the field’s understanding of the food-related physiologic processes underlying health and the prevention of disease is expanding constantly. “We are faced with the need to accumulate new information relating to how dietary patterns, specific foods, nutrients, bioactive components, and physical activity influence these processes,” she said. In addition, emerging evidence suggests that many subpopulations have differential responses to diet and chronic disease risk and that the large interindividual variability and individual responses to diets and environment are not well characterized.

The increase in diet-related chronic diseases is complex and has multiple etiologies, said Davis. The field appreciates that individual, genetic, epigenetic, phenotypic, social, and microbiome differences influence how dietary intake and physical activity affect health. “Decades of human nutrition research and advances in information technology have left us with substantial amounts of data potentially relevant to human nutritional requirements, but assimilating and using these data has been problematic,”

said Davis. Recent advances in information technology, including AI/ML, now offer possibilities of searching massive and disparate datasets and integrating multidimensional data on diet, genetics, epigenetics, microbiome, environmental factors, and other factors into a coherent framework.

Science Informing Policy

Jennifer Tiller, deputy staff director for the House Committee on Agriculture, said that her work operationalizing workforce development programs showed her that sometimes policy makers and agencies with the best intentions do not always get policies right because they were not supported by data. While considering reauthorization of the Farm Bill, Congress will deliberate, debate, and draft policies that will affect every part of the agricultural value chain. The House committee chair, she said, firmly believes that policy should use the best science—not political science—and has called for improved nutrition policies that can mitigate increasing instances of diet-related chronic disease among the population served by the programs the committee authorizes.

Tiller said that the largest of these is the Supplemental Nutrition Assistance Program, which serves over 41 million people at an annual cost of $115 billion. Previous testimony before the committee explained that the right resources, research, data, modernized programming, technology, and appropriate and effective federal dietary policy will enable USDA, the states, communities, and academia to improve the nutrition of the millions of Americans who rely on this program. “Every corner of the value chain needs to ensure there is a range of tools to help individuals from all walks of life prevent and conquer instances of disease,” said Tiller.

She noted that obesity costs the nation approximately $147 billion in annual health care costs. It also affects quality of life, general longevity, and everything from employment to military readiness. “What we consume matters, and strong, scientifically rigorous federal dietary policy is important to course correct,” she said, “No more are programs under the [House] committee’s jurisdiction only about hunger. They are now about health, and I think everyone in this room welcomes that evolution.”

Something equally important to what Americans consume is educating those who consume, said Tiller. Millions of low-income families participate in a range of nutrition education programming every year. Each program has different rubrics to capture data and measure outcomes, resulting in a questionable effect on those who need this information and education the most. “There exists a critical need for common metrics and an evaluation framework that allows the agencies with oversight of these important programs to house a repository of data that can change our programming for the better,” said Tiller. AI/ML have the potential to streamline and synthe-

size scientific advances that can increase the credibility and transparency of dietary guidance to improve the health of our nation.

ORGANIZATION OF THE PROCEEDINGS

This Proceedings of a Workshop summarizes the presentations. The speakers, panelists, and participants presented a broad range of views and ideas. Following this introductory chapter, Chapter 2 summarizes three presentations that set the stage for the workshop. Chapter 3 recounts the discussions about applications of advanced computation, big data analytics, and high-performance computing and lessons learned. Chapters 4 and 5 report on the discussions about capacity building and potential AI applications to large-scale food and nutrition initiatives, respectively. The final chapter presents a synthesis of the workshop’s key ideas to move the field forward. Appendixes A and B contain the agenda and biographical sketches of the speakers and session moderators, respectively. The speakers’ presentations (as PDF and video files) have been archived.²

___________________

² Available at https://www.nationalacademies.org/event/40460_10-2023_the-role-of-advanced-computation-predictive-technologies-and-big-data-analytics-related-to-food-and-nutrition-research-a-workshop (accessed January 9, 2024).