MIT researchers want to build database of human excrement and are asking people to send them photos of their bowel movements for analysis
- Researchers want 100,000 photos of feces to train an AI
- The project is run by Auggi, an MIT startup, and Seed, a microbiome company
- The team puts the photos into one of seven types using the Bistol Stool Scale
There are 79 million people around the world defecating at any given time of day.
A new technology company hopes that just 100,000 of them will be willing to turn around and take a picture after they’ve finished, and then send it to them for research.
Seed, a collective that researches the human microbiome, has partnered with researchers from MIT who are working on an artificial intelligence called Auggi that’s capable of analyzing images of human feces.
‘Every day, you flush away a goldmine of data—your poop’s size, shape, color, texture, consistency, and frequency can offer important insights in your overall health,’ the group’s website claims.
Scroll down for video
A team of gut researchers from MIT and Seed are hoping to get 100,000 feces photos to help build a database that shows general gut health
Gut conditions are a common yet taken-for-granted phenomenon that the team hope to learn more about through their research.
They estimate as many as three-quarters of the world population suffer from some form of digestive issue, including constipation, diarrhea, bloating, and irregular bowel movements.
By training Auggi to learn how to categorize user photos using the Bristol Stool Scale in order to begin building a generalized database of human feces in order to make inferences about overall bowel health.
People interested in participating can go to seed.com/poop to sign up. The team promises all photos will be anonymized before being added to the database
The Bristol Stool Scale ranges from Type 1, defined as ‘separate hard lumps’ associated with constipation, to Type 7, which is ‘liquid consistency with no solid pieces’ and often associated with gut inflammation.
Normal stool is associated with either Type 3, ‘a sausage shape with cracks in the surface,’ or Type 4. ‘like a smooth, soft sausage or snake.’
To ensure users’ privacy is protected, the researchers will strip out all metadata and other identifying information attached to the photo before sending it to Auggi for analysis.
To begin with, the team made prototypes using Play-Doh to test Auggi’s general shape recognition capability.
‘We spent countless hours just making different Play-Doh models,’ Auggi co-founder David Hachuel told The Verge.
‘We actually 3D-printed a toilet just to emulate how that would happen in real life.’
Later the group turned to Reddit, where they found a substantial number of users already posting photos of their own bowel movements.
The Bristol Stool Scale was formalized in 1992 and ranks stool into seven different types
The long-term goal of the project is to create a tool that will let users better manage their own gut health, especially people struggling with digestive issues.
People interested in participating can go to seed.com/poop.
The page will load on desktop computers but has been optimized for mobile phones.