Wisdom Tooth

Inspiration

In an attempt to address an overly common, but far too overlooked issue, we set out to alleviate stress and boost performance for the doctors with the highest suicide rates: dentists. By eliminating the constant stress of tedious note taking and impromptu diagnosis dental workers face on a daily basis, the mother of one of our teammates (a dentist) attested to the value Wisdom Tooth provides.

What it does

Using a convolutional neural network model, our application predicts the type of dental issue a person may have based on x-rays. By collecting and storing user data, we make personalized calls to the OpenAI API, providing in depth analysis and feedback.

How we built it

We began by training our machine learning model to detect dental issues with above a 77% accuracy rate and validated it using publicly available diagnoses we were studying. Next, we implemented a prediction algorithm to apply the learning model to make probable diagnosis on user inputted information. With a generated diagnosis, we created a full stack application using Node.js and stored user data with MongoDB. Finally, we combined user data and our predicted diagnosis to make a personalized call to the OpenAI API to further analyze possible treatments and outcomes.

Challenges we ran into

Having explored different avenues of the project at various times, the main obstacle our team faced was synthesizing several code segments. Integrating the machine learning model with the full stack application developed in parallel, in particular, posed an especially hard challenge. We spent several hours rewriting code to work injunction with each other, which ate many hours that could have gone into developing a more appealing frontend.

Accomplishments that we're proud of

Writing a machine learning model that is trained to predict the diagnosis of a certain dental issue, based on an inputted x-ray scan. Utilizing MongoDB, we were able to set up a sign-up and login page for users to enter and store their personal information such as age, medical records, allergies, and medication. This information was then used in a ChatGPT API call, to present the user with a formal dental note.

What we learned

Throughout this project we had different things working in the backend and the frontend, so this was a fullstack project. This experience taught us how to write applications from front to back and showed us how every part works together. The basis of our project, the image classification model, was tough to create and it allowed us to really learn how a machine learning model works and what it takes to create one.

What's next for Wisdom Tooth

Our next goal as a group is to improve upon Wisdom Tooth, focusing on seamless integration and an attractive frontend. Ideally, we would like to publicly release our application under a custom domain name, making it available to dentists worldwide.