By Michael J. Critelli | MakeUsWell Newsletter, 

Recently, I spoke with a nutritionist about our food intelligence product. We began talking about food allergies and sensitivities, an issue that affects a great many people.

One example was deceptively simple: someone experiences digestive distress after eating a sandwich made with whole-wheat bread.

The natural conclusion might be: Something about wheat doesn't agree with me. That could lead to questions about celiac disease, wheat allergy or non-celiac gluten/wheat sensitivity.

But there is another possibility many people would never think to investigate: fructans.

Fructans are fermentable carbohydrates found in wheat and many other foods. In susceptible people, they can contribute to gas, bloating and other gastrointestinal symptoms. Research has even found that among some people who believed themselves sensitive to gluten, fructans produced more symptoms than gluten itself.

Suddenly, the investigative question changes.

Instead of asking only, “Am I sensitive to wheat or gluten?” we begin asking, “What exactly did I eat, how much did I eat, what else did I eat, when did the symptoms begin, and have similar foods produced the same reaction?”

That is exactly what a good nutritionist wants to know.

A nutritionist might ask someone experiencing these symptoms to keep a detailed food diary. But the usefulness of the diary depends enormously on what the person notices and records.

“I ate a salad for lunch” isn't particularly informative.

What was in the salad? Onions? Garlic? Dried fruit? Cheese? Which dressing? How much? What did you drink with it? When did symptoms begin?

Those seemingly insignificant details can change the investigation completely.

The same is true of bread. Whole-wheat, rye, conventional white bread and traditionally fermented sourdough can differ significantly in their fermentable carbohydrate content. Traditional sourdough fermentation can reduce fructan levels, although the result depends on the flour and fermentation process.

Most of us don't record the type of bread we ate. We simply record “sandwich.”

But the missing detail may be the clue.

This is where I began to see a much more important role for our food intelligence product.

It isn't simply answering questions about whether a food is “good” or “bad.” It can help someone conduct a better investigation.

I thought about my own experiences.

Sugar-free chewing gum has sometimes caused digestive distress for me. Many sugar-free gums and mints contain sugar alcohols such as xylitol or sorbitol, which can produce gastrointestinal symptoms in some people. A food intelligence system can prompt someone not simply to record “chewing gum,” but to look at the ingredients and notice which sweetener was present.

My experience with dairy provides another example.

At about age 30, a physician told me I was lactose intolerant after I experienced digestive distress from drinking milk. He told me to steer clear of all dairy products. Yet I could tolerate many cheeses and even some frozen dairy desserts.

At first, those experiences seemed inconsistent.

They aren't necessarily.

Lactose intolerance is not usually an all-or-nothing phenomenon. Most people who have it can tolerate some lactose, and hard cheeses such as cheddar and Swiss generally contain considerably less lactose than milk. Amount, food type and what else is consumed can all affect the outcome.

The lesson isn't that an AI food product should diagnose lactose intolerance, fructan sensitivity or anything else.

The lesson is that it can help us discover what questions should be asked.

Imagine experiencing digestive distress several times over a month. Without assistance, you might remember three unrelated episodes: a sandwich, a restaurant salad and a snack containing sugar-free candy.

There may be no obvious pattern.

A food intelligence system can help you reconstruct those episodes, examine ingredients, quantities and timing, identify possible common denominators, and suggest information worth capturing the next time the problem occurs.

It can then help you bring a much richer record to a nutritionist or physician.

That distinction is important.

Physicians, physician assistants, nurses and nutrition professionals possess expertise their patients generally do not. But they also work under significant time constraints. They cannot analyze information we never noticed, never recorded or never thought important enough to mention.

AI can help close that gap.

The most valuable contribution of our food intelligence product may therefore be neither diagnosing disease nor prescribing diets. It may be something more fundamental: teaching us what to notice.

It can prompt us to capture the ingredient we would otherwise overlook, the portion size we would forget, the timing we would dismiss, or the pattern across several meals that we would never connect.

We still need qualified healthcare and nutrition professionals to interpret those clues and decide what they mean.

But their investigation becomes much more powerful when we arrive with better evidence.

That may turn out to be one of AI's most important roles in healthcare: not replacing professional judgment, but giving professionals much better information on which to exercise it.