Stool Tracker

A quick and easy way to track and identify different bowel movements using AI. Just take a quick picture and the app will analyze and store your stool!

Comment

Inspiration

Bowel movements are often seen as mundane tasks or a second-thought for many people. We wanted to change that by showing the importance of tracking stool progress and how it can actually greatly affect a person's health. We referred to pre-existing stool tracking apps and utilized user feedback like cleaner UI, giving reliable information, and making it not restricted for all ages (some apps are only for children).

What it does

The Stool Tracker prompts the user to snap a photo of their stool and using our trained AI model, it determines whether the user has a healthy stool or a bad stool. Then it bunches up their stool history to provide an accurate tip overview where the model gives a rating from either constipation to diarrhea. Then it records the number of times the user has experienced bowel movement and adds it to the calendar so they are able to see how many stools they've produced within the week. Finally, it produces a tailored tip of the day/week based off the user's stool quality, such as if it was a good stool then there is only praise but if there is constipation then the AI would recommend more fiber intake and to maintain hydration.

How we built it

We first trained an AI model to recognize different stool levels based off the Bristol Stool Chart. Then we started prototyping on Figma to get a general idea of features, UI, as well as how convenient navigation would be. Then with the frontend we used typescript and then for backend it was python. Then we used Xcode to get a visual representation of what the frontend would look like on an Iphone 16 mobile device.

Challenges we ran into

Cleaning and labelling a dataset of real human stool is no easy feat. Also considering there were some abnormalities in the photos that would hinder the results of the AI's recognition abilities.

Accomplishments that we're proud of

Being beginner hackers with not much experience, we are proud that we were able to create something

What we learned

We learned a lot of things since we are beginner hackers but I think our biggest lesson was learning how to train an AI model since neither of us have ever done that before so it was pretty cool to see it working in our app.

What's next for Poop Tracker

  • Continue improving the model performance, especially when the photo has extra objects (ex. toilet paper).
  • Improve tips toolbar by analyzing bowel movements over multiple days, rather than just yesterday.
  • Add a sign up/log in page to create a more personalized user experience.
  • Hopefully more healthy poops :)
Share this project:

Updates