by Mike Critelli, 

Today, we are discussing one big health management issue. As consumers of foods, beverages, over-the-counter medications, and a wide range of supplements, how can AI aid in the process of ensuring that we are putting the right things into our bodies at the right time and in the right ways?  

The first priority is to make sure that, whether we are using human intelligence, AI, or a combination of both, we are using authoritative sources. Any AI platform focused on health-related advice also has to be based firmly on authoritative, unassailable sources. Uncontrolled use of, or faith in, AI can easily lead users to sources with credible, but wrong or unsupported, information. Researchers are held to a high standard in peer-reviewed articles. AI’s output must draw on the same sources and be held to the same standards as humans.

However an AI platform might find a source, it must present the same transparency and rigor as a research paper or book that provides auditable sites for every source. Someone who wants to know the source or sources of the information an AI agent provides must be able to query it and get back to the primary source of the information.  

AI can not only disclose the sources used, it can also be used to detect and disclose authoritative sources that indicate potential disagreements among authorities. 

Any AI-powered resource we provide will also have five other vital attributes:

• Individuals will find it easy to use. At the same time, it will always aim to educate individuals about the trade-offs and complexities of self management of health. 

• It should caution users that, particularly with respect to supplements, online purchases carry the risk of receiving an item in which the contents do not match what the purchaser believes he or she is ordering.  

• It must remind users they should consult with trained healthcare professionals with relevant domain knowledge and with an understanding of how their medical condition and history can alter a decision. AI can draw on ongoing research relative to the most effective disclosures relative to the risks and benefits of medications and supplements and the best way to guide people to healthcare professionals.

The conclusions AI yields will help us understand that the more complexity food additives, medications and supplements introduce into our daily lives, the more potential there is for unintended consequences. AI should also help to educate us that newly diagnosed medical conditions, such as autoimmune diseases, can alter dietary decisions, as can the progression of women through different stages of life, such as menopause.  

In the US, the inclusion of additives to foods and beverages, and the marketing of supplements are poorly regulated or not regulated at all. The FDA does not require disclosure of many kinds of food additives. Unlike the EU, which enforces a “precautionary principle” that requires an upfront proof of no harm, the US’s “generally regarded as safe” standard is far more permissive. 

Consumers should have access to EU research and findings when they are available and relevant. There are two reasons for this: first, as is done when EU models are cited in alternative weather reports to track the potential paths of forthcoming hurricanes, authoritative sources may draw different conclusions from the same data. Second, AI needs to signal to users that, depending on their tolerance of risk, they may make different choices.  

Our goal is not to replace the role of healthcare professionals, pharmacists or other resources familiar with an individual’s health profile. It is to help consumers understand when and how they might need help and how to collect data and frame questions that yield better insights. Most importantly, it is to make them better able to manage the myriad decisions they must make in their lives outside the healthcare system.

Neither human intelligence nor AI can easily and with certainty yield firm conclusions in many situations. Moreover, authoritative clinical studies often inform us that decisions we make relative to foods, medications and supplements to address a specific medical condition may reduce or harm our ability to address a different one. This is particularly important because so many Americans suffer from multiple chronic conditions.  

AI may also yield conclusions that induce scientists to engage in new research. We would make what’s available as concise, accessible, understandable, actionable and timely as possible. Our goal is to bring better insights to users than what they would access unaided.  

The most valuable insights we can deliver through AI would be those that help users appreciate that health management is a complex, continual process that they need to own through good habits they build into their daily lives. We should be cautious toward any claims of quick-fix health solutions and know what we are doing before we start adding any medications or supplements into our bodies.