Here’s an opportunity for the charitable scientific enthusiasts among us: please photograph your poop today in order to help train artificial intelligence.
Poops of all shapes and sizes are welcome, for the microbial sciences company Seed will use them to create a database of 100,000 prospective images based on shape and consistency. A team of gastroenterologists will then sort each image on the Bristol scale (seven categories of poop consistencies) from hard nut-like clumps (indicating constipation) to liquid (indicating diarrhoea). A new AI platform dubbed auggi–which has been trained to identify poops via 36,000 images of Play-Doh models–will then learn to distinguish between healthy and unhealthy poop.
A poop identification tool would serve an important social service, as some research estimates that roughly one in five Americans suffer from gastrointestinal conditions. #GiveaShit hopes to de-stigmatised poop, and based on the Twitter results for #GiveaShit, netizens are on the job.
Poop consistency might be self-evident to poopers, but the pattern matters; at the moment, doctors are left to patients’ eyeballing, which leaves diagnoses to guesswork. David Hachuel, CEO and co-founder of auggie compared it to studying heart conditions via heartbeats: one nutty poop may be an anomaly, but repeated nutty poops might indicate chronic constipation. Other poop consistencies could indicate potential signs of irritable bowel syndrome or Crohn’s disease.
Ara Katz, co-founder and co-CEO of Seed, hopes that the poop project is just one of the company’s many future contributions to our understanding of health. “It’s projects like this [that] allow people who are not scientists to participate in citizen science. By crowdsourcing data, we can help researchers and technologies like auggi in order to help people identify different conditions.”
So #GiveaShit before you flush. You can photograph your poop and submit it via your phone at seed.com/poop. You’ll have to enter your email, but Seed ensures us that your image will not be associated with your personal information.