Two Startups Are Training AI to Identify Your Poop
Thu, April 22, 2021

Two Startups Are Training AI to Identify Your Poop

Startup companies are developing apps to help people monitor their gastrointestinal maladies / Photo Credit: DestroLove (via Shutterstock)


Startups are going to snap photos of your droppings for scientific purposes, according to Rachel Metz of American news channel CNN Business. Gut-health startup Auggi is developing an app for people to track gastrointestinal issues. Likewise, Seed Health is working on “applying microbes to human health and sells probiotics.” This startup has been soliciting photos of droppings from people who want to send them. 

The two companies have been collecting photos as they want to create the first data set of human poop images. With the hope of collecting 100,000 photos, these pictures can be used to build an AI for research into gastrointestinal maladies. The AI will also help people track their own bowel movements. Seed co-founder and co-CEO Ara Katz told CNN Business, “We like to say it's basically a data dump that gets flushed away each day that could really inform science.” 

Auggi and Seed Health believe that many people would benefit from their research. For instance, Auggi hopes to use the photos of poop to develop an app “that can use computer vision” to automatically classify various types of stools that individuals with chronic gut problems have to manually track using a pen and paper. Auggi co-founder and CEO David Hachuel hopes to release its app in the first quarter of 2020. 

Patients suffering from gastrointestinal conditions are made to keep a log of the properties of their stools using the Bristol tool chart. The chart is used by medical professionals to classify stool based on their consistency. According to professor of pediatrics at the University of California San Diego of Medicine and co-founder of American Gut Project Jack Gilbert, patients are required to rate their stool using the chart in every “clinical trial he conducts.”

However, automating the said process helps minimize bias and data variation. Auggi expressed its interest in automating this process. Photos will be analyzed by a team of gastroenterologists, who will classify each stool according to the Bristol Chart, Hachuel said. Then, the images would be processed by a computer. It would be trained to spot the differences between various categories of stool.  

Hachuel added, “Obviously, when it comes to real stool, we need real data to achieve similar levels of accuracy.”