Would you expect a single set of clothes to fit everyone? Of course not. Similarly, would you expect a single diet to work for everyone? Again, that's not realistic. People differ significantly when it comes to their surroundings, medical history, daily activities, body composition, biology, genetics, and the many other factors affecting diet and health.
Society needs to move away from the one-diet-fits-all approach that's been used for years and toward precision nutrition, an emerging field that involves tailoring diets so they better match different people's circumstances and characteristics. Computer technology and artificial intelligence (AI) will be central to achieving more widespread precision nutrition.
Why Do We Need AI for Precision Nutrition?
A complex system of genetic, biological, behavioral, social, environmental, and economic factors and mechanisms affects and is affected by one's diet and health. A patient's metabolism can be influenced by their eating behaviors, sleep patterns, and stress levels, which in turn can be affected by their work, environment, and social circles. Their risk of developing cardiovascular disease can be related to not only what they eat but also their genetics, physical activity, living situation, comorbid medical conditions, and a host of other factors.
Humans can typically figure out simple direct cause-and-effect relationships, but we tend to struggle when they become more complex. Computer-aided approaches like AI could transform nutrition by helping make sense of large amounts of information and patterns that are too complex for most people to understand.
However, the impact of AI will depend on how it is actually used. When used in the right manner, AI could be a positive game changer, helping to make nutrition guidelines and recommendations much more precise and individually tailored in ways that improve diet and health. When used inappropriately, however, it could make things worse, leading to inappropriate recommendations, the introduction of bias, and the worsening of already existing disparities in nutrition and health.
How Would the AI Be Used?
While it's recently become increasingly fashionable to use, AI is a very broad and vague term encompassing a wide range of approaches and techniques. The Oxford Dictionary defines AI as "the theory and development of computer systems able to perform tasks normally requiring human intelligence, such as visual perception, speech recognition, decision-making, and translation between languages."
Therefore, saying that you are using AI to tackle a problem is sort of like saying you are using computers to tackle a problem. Or that you are going to use medicine to address a health issue. That alone doesn't say much.
So, when you hear that AI is being used, dig deeper. Ask how it is specifically being used. How exactly is AI helping disentangle the complex systems involved?
Along with Diana Thomas, PhD, a professor of mathematics at West Point, I co-lead the new AIMINGS (Artificial Intelligence, Modeling, and Informatics, for Nutrition Guidance and Systems) Center that will serve as the AI Center for Precision Nutrition and Health. Our charge is to develop new AI algorithms to help better tailor diets to different people's circumstances and characteristics. To do this, we will be incorporating the principles and practices of systems science, an interdisciplinary field that entails developing and implementing approaches and methods to better understand and address complex systems.
What Might This Look Like in Practice?
Let's say you want to develop an AI-based app that helps patients lose weight. If this app simply focused on a patient's current height and weight, their microbiome, and a few other measures from blood tests such as their blood sugar, it could offer impractical advice or even the wrong conclusions. After all, it would be completely ignoring the complex systems outside their body and in their surroundings. There's a big difference between someone living a comfortable and already otherwise healthy life in a mansion with a personal chef and a fitness trainer versus someone who is working three different stressful jobs, living in a high crime neighborhood, and unable to afford fresh fruits and vegetables.
You'd also want to know specifically how the app is characterizing the relationship between a particular factor, such as the composition of the patient's microbiome, and how they may be able to lose weight. Does the algorithm oversimplify this relationship? Did the developers of the app simply observe a small group of people and look for simple correlations? Were those people similar to your patient and in what way? Trying to generalize relationships found within other people, who may be very different than your patient, would in essence be trying to force a one-size-fits-all approach in a different way.
We are at a key inflection point in society. Years of research have helped scientists realize the complexity of the relationships between nutrition and health. Meanwhile, we now have more computer-based approaches and much more information available, with so many people using wearables, the Internet, and other technologies. Diet and nutrition apps have already started to emerge. But it's unclear how many of these are really useful and how many may actually be misleading.
AI certainly has the potential to transform nutrition and health, depending on how it is used. The question is whether this transformation will be positive or negative.
Bruce Y. Lee, MD, MBA, is a professor of health policy and management at the City University of New York (CUNY) Graduate School of Public Health and Health Policy; executive director of Public Health Informatics, Computational, and Operations Research (PHICOR) and the Center for Advanced Technology and Communication in Health (CATCH); co-PI of the AIMINGS Center; and founder and CEO of Symsilico.
He is internationally recognized for his work developing AI, computational models, and other computer-based approaches to help health-related decision making. He also writes extensively for the general media, including covering health and healthcare as a senior contributor for Forbes and maintaining his "A Funny Bone to Pick" blog as a regular contributor for Psychology Today.