Written by: Asaf Kraus, Founder, Dieta Health
Stool Scan: the first clinically validated stool imaging AI, now built into a virtual-first digestive care program. I want to share the story behind it and why I think it can meaningfully change how digestive conditions are understood and managed.
A patient experience shared by millions
Back in 2017, I was working as a data scientist at Uber in San Francisco when I got hit with severe IBS. Within a few months it had taken over my life.
I spent the better part of a year visiting gastroenterologists, trying dozens of interventions, and getting inconsistent recommendations. Nobody could tell me clearly what I had or what would actually help.
That experience is incredibly common. Millions of Americans experience digestive conditions each year, and many go through the same cycle of confusing guidance, trial and error, and a system that doesn’t do a great job capturing what’s actually happening in their bodies.
That disconnect is what eventually led to Stool Scan. But the moment it really clicked was much simpler than that.
A poster on the wall
Every gastroenterologist’s office has the same poster hanging on the wall: the Bristol Stool Chart. Seven cartoon poops numbered 1 through 7, from hard lumps to liquid.
The doctor asks you to describe how your stool looks and how often each type shows up.
The problem is those cartoon diagrams don’t really match what people actually see. Stool isn’t one variable. It’s not just “type 4” or “type 6.”
Is it one piece or many?
Is it large or small?
Liquid or solid?
How difficult was it to pass?
The Bristol chart lumps multiple characteristics together into a single number and then asks the patient (not a clinician) to make that judgment call from memory.
That’s a lot of cognitive burden for something that’s already uncomfortable to think about—and easy to get wrong.
Then it got worse
Six months later, out of desperation, I enrolled in a clinical trial for an experimental IBS medication. That’s when I realized the stool description problem was even bigger than I thought.
The way this pharmaceutical company measured whether their drug worked was by asking patients to rate their own stool on that same Bristol scale.
That was the central outcome measure.
The data going to the FDA to determine whether this drug should exist in the market was based on patients’ subjective perception of their own stool.
The same problem everywhere
This wasn’t just a problem in the doctor’s office.
This is how gastroenterologists deliver care.
This is how GI researchers evaluate new treatments.
This is how food companies and supplement companies measure results.
And all of it is subjective, prone to error, dependent on patient recall, and limited to a single variable.
At Uber, I had spent my days surrounded by people who were obsessed with capturing better data to personalize experiences at scale. I looked at this system and thought: this could be dramatically improved with better data and better technology.
What we built
That’s what we set out to do at Dieta Health — to build the first mobile app with stool imaging AI.
Instead of asking patients to look at their stool, compare it to a cartoon, and pick a number — they just point their smartphone and take a photo. The AI does the analysis.
It doesn’t just replicate the Bristol chart. It captures multiple variables simultaneously: consistency, fragmentation, volume, color, blood or mucus. Patients don’t have to think about any of it. They take a picture, and the AI does the thinking and the remembering.
We published four clinical trials validating this technology with researchers at Cedars-Sinai and Mayo Clinic.
The key finding was simple:
When you compare patient self-assessment to our AI against a gastroenterologist’s expert evaluation as the source of truth, the AI is more accurate than the patient. And it captures significantly more information per bowel movement.
Over the next few years, 25,000 patients used the app, collecting over half a million annotated stool images. We showed that better stool data improves patient stratification. We could identify subgroups within an IBS population that respond differently to the same treatment.
That’s the kind of insight that helps a clinician choose the right intervention for a specific patient instead of guessing.
Why this matters inside Cylinder
In 2025, Cylinder Health acquired Dieta, and our team joined the company. That changed the scale of what we could do.
We had strong technology and strong validation, but we were a small startup. Now this technology lives inside a digestive health center of excellence, with over 145,000 members and a full clinical team of gastroenterologists, registered dietitians, and health coaches.
Stool data is no longer floating in isolation.
It’s reviewed alongside food intake, symptom tracking, behavioral context, and real clinical interactions, to create a more complete clinical picture.
That’s when stool imaging goes from being a measurement tool to being a real driver of better care. For example, when a patient starts a low-FODMAP diet and photographs their stool for a few weeks, we can show them their stool became measurably more consistent. That’s the kind of biofeedback that motivates people to stick with difficult interventions. And when our clinical team can see those patterns changing, or worsening, they can act earlier instead of waiting for the next appointment.
What’s ahead
What excites me most is everything we haven’t built yet.
With Cylinder’s patient scale, we can grow our dataset dramatically, add new stool variables, and make the AI more accurate across a wider range of conditions.
We can build automatic alerts for the care team when stool patterns shift, when blood appears, or when trends suggest a flare or something more serious.
We can cluster patients into more precise subgroups and use that to inform treatment recommendations,getting closer to truly personalized GI care instead of the guess-and-check process that millions of patients endure today.
Stool Scan launching to all members is the first step. We’re just getting started, and the opportunity in front of us is massive.
We started this work because I was a frustrated IBS patient who happened to be a data scientist.
I still believe the same thing I believed then: better data leads to better care. Now we have the technology, the validation, and the scale to make a real impact on patients’ lives.
