Which foods are actually healthy for your individual gut microbiome? Several companies are offering personalized dietary guidance based on your test results, but their answers in one experiment turned up with some conflicting advice.
I recently got on the scale to weigh myself, thinking I've got to eat better. With so many trendy diets today claiming to improve health, from Keto to Paleo to Whole30, it can be confusing to figure out what we should and shouldn't eat for optimal nutrition.
A number of companies are now selling the concept of "personalized" nutrition based on the genetic makeup of your individual gut bugs.
My next thought was: I've got to lose a few pounds.
Consider a weird factoid: In addition to my fat, skin, bone and muscle, I'm carrying around two or three pounds of straight-up bacteria. Like you, I am the host to trillions of micro-organisms that live in my gut and are collectively known as my microbiome. An explosion of research has occurred in the last decade to try to understand exactly how these microbial populations, which are unique to each of us, may influence our overall health and potentially even our brains and behavior.
Lots of mysteries still remain, but it is established that these "bugs" are crucial to keeping our body running smoothly, performing functions like stimulating the immune system, synthesizing important vitamins, and aiding digestion. The field of microbiome science is evolving rapidly, and a number of companies are now selling the concept of "personalized" nutrition based on the genetic makeup of your individual gut bugs. The two leading players are Viome and DayTwo, but the landscape includes the newly launched startup Onegevity Health and others like Thryve, which offers customized probiotic supplements in addition to dietary recommendations.
The idea has immediate appeal – if science could tell you exactly what to make for lunch and what to avoid, you could forget about the fad diets and go with your own bespoke food pyramid. Wondering if the promise might be too good to be true, I decided to perform my own experiment.
Last fall, I sent the identical fecal sample to both Viome (I paid $425, but the price has since dropped to $299) and DayTwo ($349). A couple of months later, both reports finally arrived, and I eagerly opened each app to compare their recommendations.
First, I examined my results from Viome, which was founded in 2016 in Cupertino, Calif., and declares without irony on its website that "conflicting food advice is now obsolete."
I learned I have "average" metabolic fitness and "average" inflammatory activity in my gut, which are scores that the company defines based on a proprietary algorithm. But I have "low" microbial richness, with only 62 active species of bacteria identified in my sample, compared with the mean of 157 in their test population. I also received a list of the specific species in my gut, with names like Lactococcus and Romboutsia.
But none of it meant anything to me without actionable food advice, so I clicked through to the Recommendations page and found a list of My Superfoods (cranberry, garlic, kale, salmon, turmeric, watermelon, and bone broth) and My Foods to Avoid (chickpeas, kombucha, lentils, and rice noodles). There was also a searchable database of many foods that had been categorized for me, like "bell pepper; minimize" and "beef; enjoy."
"I just don't think sufficient data is yet available to make reliable personalized dietary recommendations based on one's microbiome."
Next, I looked at my results from DayTwo, which was founded in 2015 from research out of the Weizmann Institute of Science in Israel, and whose pitch to consumers is, "Blood sugar made easy. The algorithm diet personalized to you."
This app had some notable differences. There was no result about my metabolic fitness, microbial richness, or list of the species in my sample. There was also no list of superfoods or foods to avoid. Instead, the app encouraged me to build a meal by searching for foods in their database and combining them in beneficial ways for my blood sugar. Two slices of whole wheat bread received a score of 2.7 out of 10 ("Avoid"), but if combined with one cup of large curd cottage cheese, the score improved to 6.8 ("Limit"), and if I added two hard-boiled eggs, the score went up to 7.5 ("Good").
Perusing my list of foods with "Excellent" scores, I noticed some troubling conflicts with the other app. Lentils, which had been a no-no according to Viome, received high marks from DayTwo. Ditto for Kombucha. My purported superfood of cranberry received low marks. Almonds got an almost perfect score (9.7) while Viome told me to minimize them. I found similarly contradictory advice for foods I regularly eat, including navel oranges, peanuts, pork, and beets.
Contradictory dietary guidance that Kira Peikoff received from Viome (left) and DayTwo from an identical sample.
To be sure, there was some overlap. Both apps agreed on rice noodles (bad), chickpeas (bad), honey (bad), carrots (good), and avocado (good), among other foods.
But still, I was left scratching my head. Which set of recommendations should I trust, if either? And what did my results mean for the accuracy of this nascent field?
I called a couple of experts to find out.
"I have worked on the microbiome and nutrition for the last 20 years and I would be absolutely incapable of finding you evidence in the scientific literature that lentils have a detrimental effect based on the microbiome," said Dr. Jens Walter, an Associate Professor and chair for Nutrition, Microbes, and Gastrointestinal Health at the University of Alberta. "I just don't think sufficient data is yet available to make reliable personalized dietary recommendations based on one's microbiome. And even if they would have proprietary algorithms, at least one of them is not doing it right."
There is definite potential for personalized nutrition based on the microbiome, he said, but first, predictive models must be built and standardized, then linked to clinical endpoints, and tested in a large sample of healthy volunteers in order to enable extrapolations for the general population.
"It is mindboggling what you would need to do to make this work," he observed. "There are probably hundreds of relevant dietary compounds, then the microbiome has at least a hundred relevant species with a hundred or more relevant genes each, then you'd have to put all this together with relevant clinical outcomes. And there's a hundred-fold variation in that information between individuals."
However, Walter did acknowledge that the companies might be basing their algorithms on proprietary data that could potentially connect all the dots. I reached out to them to find out.
Amir Golan, the Chief Commercial Officer of DayTwo, told me, "It's important to emphasize this is a prediction, as the microbiome field is in a very early stage of research." But he added, "I believe we are the only company that has very solid science published in top journals and we can bring very actionable evidence and benefit to our uses."
He was referring to pioneering work out of the Weizmann Institute that was published in 2015 in the journal Cell, which logged the glycemic responses of 800 people in response to nearly 50,000 meals; adding information about the subjects' microbiomes enabled more accurate glycemic response predictions. Since then, Golan said, additional trials have been conducted, most recently with the Mayo Clinic, to duplicate the results, and other studies are ongoing whose results have not yet been published.
He also pointed out that the microbiome was merely one component that goes into building a client's profile, in addition to medical records, including blood glucose levels. (I provided my HbA1c levels, a measure of average blood sugar over the previous several months.)
"We are not saying we want to improve your gut microbiome. We provide a dynamic tool to help guide what you should eat to control your blood sugar and think about combinations," he said. "If you eat one thing, or with another, it will affect you in a different way."
Viome acknowledged that the two companies are taking very different approaches.
"DayTwo is primarily focused on the glycemic response," Naveen Jain, the CEO, told me. "If you can only eat butter for rest of your life, you will have no glycemic response but will probably die of a heart attack." He laughed. "Whereas we came from very different angle – what is happening inside the gut at a microbial level? When you eat food like spinach, how will that be metabolized in the gut? Will it produce the nutrients you need or cause inflammation?"
He said his team studied 1000 people who were on continuous glucose monitoring and fed them 45,000 meals, then built a proprietary data prediction model, looking at which microbes existed and how they actively broke down the food.
Jain pointed out that DayTwo sequences the DNA of the microbes, while Viome sequences the RNA – the active expression of DNA. That difference, in his opinion, is key to making accurate predictions.
"DNA is extremely stable, so when you eat any food and measure the DNA [in a fecal sample], you get all these false positives--you get DNA from plant food and meat, and you have no idea if those organisms are dead and simply transient, or actually exist. With RNA, you see what is actually alive in the gut."
More contradictory food advice from Viome (left) and DayTwo.
Note that controversy exists over how it is possible with a fecal sample to effectively measure RNA, which degrades within minutes, though Jain said that his company has the technology to keep RNA stable for fourteen days.
Viome's approach, Jain maintains, is 90 percent accurate, based on as-yet unpublished data; a patent was filed just last week. DayTwo's approach is 66 percent accurate according to the latest published research.
Natasha Haskey, a registered dietician and doctoral student conducting research in the field of microbiome science and nutrition, is skeptical of both companies. "We can make broad statements, like eat more fruits and vegetables and fiber, but when it comes to specific foods, the science is just not there yet," she said. "I think there is a future, and we will be doing that someday, but not yet. Maybe we will be closer in ten years."
Professor Walter wholeheartedly agrees with Haskey, and suggested that if people want to eat a gut-healthy diet, they should focus on beneficial oils, fruits and vegetables, fish, a variety of whole grains, poultry and beans, and limit red meat and cheese, as well as avoid processed meats.
"These services are far over the tips of their science skis," Arthur Caplan, the founding head of New York University's Division of Medical Ethics, said in an email. "We simply don't know enough about the gut microbiome, its fluctuations and variability from person to person to support general [direct-to-consumer] testing. This is simply premature. We need standards for accuracy, specificity, and sensitivity, plus mandatory competent counseling for all such testing. They don't exist. Neither should DTC testing—yet."
Meanwhile, it's time for lunch. I close out my Viome and DayTwo apps and head to the kitchen to prepare a peanut butter sandwich. My gut tells me I'll be just fine.
Giving robots self-awareness as they move through space - and maybe even providing them with gene-like methods for storing rules of behavior - could be important steps toward creating more intelligent machines.
Now roboticists are going a step further, training their creations to do even better: understand their own image in space and interact with humans like humans do. Today, there are already robot-teachers like KeeKo, robot-pets like Moffin, robot-babysitters like iPal, and robotic companions for the elderly like Pepper.
But even these reasonably intelligent creations still have huge limitations, some scientists think. “There are niche applications for the current generations of robots,” says professor Anthony Zador at Cold Spring Harbor Laboratory—but they are not “generalists” who can do varied tasks all on their own, as they mostly lack the abilities to improvise, make decisions based on a multitude of facts or emotions, and adjust to rapidly changing circumstances. “We don’t have general purpose robots that can interact with the world. We’re ages away from that.”
Robotic spatial self-awareness – the achievement by the team at Columbia – is an important step toward creating more intelligent machines. Hod Lipson, professor of mechanical engineering who runs the Columbia lab, says that future robots will need this ability to assist humans better. Knowing how you look and where in space your parts are, decreases the need for human oversight. It also helps the robot to detect and compensate for damage and keep up with its own wear-and-tear. And it allows robots to realize when something is wrong with them or their parts. “We want our robots to learn and continue to grow their minds and bodies on their own,” Chen says. That’s what Zador wants too—and on a much grander level. “I want a robot who can drive my car, take my dog for a walk and have a conversation with me.”
Columbia scientists have trained a robot to become aware of its own "body," so it can map the right path to touch a ball without running into an obstacle, in this case a square.
Jane Nisselson and Yinuo Qin/ Columbia Engineering
Today’s technological advances are making some of these leaps of progress possible. One of them is the so-called Deep Learning—a method that trains artificial intelligence systems to learn and use information similar to how humans do it. Described as a machine learning method based on neural network architectures with multiple layers of processing units, Deep Learning has been used to successfully teach machines to recognize images, understand speech and even write text.
Trained by Google, one of these language machine learning geniuses, BERT, can finish sentences. Another one called GPT3, designed by San Francisco-based company OpenAI, can write little stories. Yet, both of them still make funny mistakes in their linguistic exercises that even a child wouldn’t. According to a paper published by Stanford’s Center for Research on Foundational Models, BERT seems to not understand the word “not.” When asked to fill in the word after “A robin is a __” it correctly answers “bird.” But try inserting the word “not” into that sentence (“A robin is not a __”) and BERT still completes it the same way. Similarly, in one of its stories, GPT3 wrote that if you mix a spoonful of grape juice into your cranberry juice and drink the concoction, you die. It seems that robots, and artificial intelligence systems in general, are still missing some rudimentary facts of life that humans and animals grasp naturally and effortlessly.
How does one give robots a genome? Zador has an idea. We can’t really equip machines with real biological nucleotide-based genes, but we can mimic the neuronal blueprint those genes create.
It's not exactly the robots’ fault. Compared to humans, and all other organisms that have been around for thousands or millions of years, robots are very new. They are missing out on eons of evolutionary data-building. Animals and humans are born with the ability to do certain things because they are pre-wired in them. Flies know how to fly, fish knows how to swim, cats know how to meow, and babies know how to cry. Yet, flies don’t really learn to fly, fish doesn’t learn to swim, cats don’t learn to meow, and babies don’t learn to cry—they are born able to execute such behaviors because they’re preprogrammed to do so. All that happens thanks to the millions of years of evolutions wired into their respective genomes, which give rise to the brain’s neural networks responsible for these behaviors. Robots are the newbies, missing out on that trove of information, Zador argues.
A neuroscience professor who studies how brain circuitry generates various behaviors, Zador has a different approach to developing the robotic mind. Until their creators figure out a way to imbue the bots with that information, robots will remain quite limited in their abilities. Each model will only be able to do certain things it was programmed to do, but it will never go above and beyond its original code. So Zador argues that we have to start giving robots a genome.
How does one do that? Zador has an idea. We can’t really equip machines with real biological nucleotide-based genes, but we can mimic the neuronal blueprint those genes create. Genomes lay out rules for brain development. Specifically, the genome encodes blueprints for wiring up our nervous system—the details of which neurons are connected, the strength of those connections and other specs that will later hold the information learned throughout life. “Our genomes serve as blueprints for building our nervous system and these blueprints give rise to a human brain, which contains about 100 billion neurons,” Zador says.
If you think what a genome is, he explains, it is essentially a very compact and compressed form of information storage. Conceptually, genomes are similar to CliffsNotes and other study guides. When students read these short summaries, they know about what happened in a book, without actually reading that book. And that’s how we should be designing the next generation of robots if we ever want them to act like humans, Zador says. “We should give them a set of behavioral CliffsNotes, which they can then unwrap into brain-like structures.” Robots that have such brain-like structures will acquire a set of basic rules to generate basic behaviors and use them to learn more complex ones.
Currently Zador is in the process of developing algorithms that function like simple rules that generate such behaviors. “My algorithms would write these CliffsNotes, outlining how to solve a particular problem,” he explains. “And then, the neural networks will use these CliffsNotes to figure out which ones are useful and use them in their behaviors.” That’s how all living beings operate. They use the pre-programmed info from their genetics to adapt to their changing environments and learn what’s necessary to survive and thrive in these settings.
For example, a robot’s neural network could draw from CliffsNotes with “genetic” instructions for how to be aware of its own body or learn to adjust its movements. And other, different sets of CliffsNotes may imbue it with the basics of physical safety or the fundamentals of speech.
At the moment, Zador is working on algorithms that are trying to mimic neuronal blueprints for very simple organisms—such as earthworms, which have only 302 neurons and about 7000 synapses compared to the millions we have. That’s how evolution worked, too—expanding the brains from simple creatures to more complex to the Homo Sapiens. But if it took millions of years to arrive at modern humans, how long would it take scientists to forge a robot with human intelligence? That’s a billion-dollar question. Yet, Zador is optimistic. “My hypotheses is that if you can build simple organisms that can interact with the world, then the higher level functions will not be nearly as challenging as they currently are.”
