JustSmile
PittChallenge Hackathon - Fall 2023
Employing machine learning technology and telemedicine to create equitable access to dental care
Clinical Problem
Dental care is a luxury.
We found that over 76.5 million Americans lack dental insurance.
Due to the high cost of care, over 50% of Americans only go to the dentist for emergencies, resulting in $45 billion dollars of lost productivity each year.
We reached out to two dentists to learn how people can be motivated to take care of their oral health without the barrier of cost.
By speaking with Dr. Dave, I was able to understand the types of cases typically seen in the free clinic where she worked. The majority of cases she saw were for patients from underserved communities who had waited until the last moment to receive care resulting in a longer and more invasive procedure. This meant they had to take time off work to receive and recover from the procedure. Additionally, by performing an invasive procedure like a tooth extraction, patients are at a greater risk for other diseases and infections.
The biggest takeaway from this interview was learning that preventative care for oral health is very important. Teaching people to care for their oral health diseases in the early stages is vital for preventing further complications.
Dr. Anjali Dave, DDS
Dr. Mark S. Goldstein, DDS
and so, we asked ourselves...
How might we address inequitable access to dental care and information?
With less than forty-eight hours to create a solution, we started with ideation using the resources available to us. To tackle this challenge, we set out to create an app that could help patients remotely classify their oral diseases and guide them on how to take care of their condition without the need for a dentist visit. First, we researched types of oral diseases and identified an image dataset we could use to create a machine-learning model to implement in this app.
I took charge of outlining all the tasks necessary to meet our goal and assigned them to each member of the team. This allowed us to keep a visual record of everything that needed to be done and be able to complete tasks in order of dependency.
My role was to create the app interface and workflow of how a potential user would use the app.
I started by outlining the most important elements of the app and drawing out what the GUI would look like.
Then, I used Flutterflow.io to create the interface.
I also used Google Firebase to store the user data and Chat GPT's API to create a custom chatbot within the app.
GUI Mockup
Screen 1: Startup screen of the app.
Screen 2: User creates an account or logs in.
Screen 3: User is prompted for additional information.
Screen 4: User uploads three photos of various angles.
Screen 5: User has the option to upload a photo or select from gallery.
Screen 6: Our ML model screens the photos uploaded for signs of oral disease. The results are displayed.
Screen 7: User had the option to ask any follow up questions via the chatbot.
Screen 8: The chat GPT API responds to the questions asked.
Screen 9: User can track photos uploaded to check progression of disease.
Screen 10: User can connect with providers and view previous treatment plans.
Using tensorflow and teachable machine, my partners were able to create the machine learning model in Python.
I utilized Android Studio to implement the model into the app using Java.
Our code is accessible on Github here:
Awards
After 48 hours of hard work, no sleep, and tons of caffeine, we presented to a panel of judges about our idea and work.
We were awarded a prize in our track: Big Data and Machine Learning
Overall, this was my first experience competing in a hackathon and I was able to learn a lot in a short amount of time. I attended workshops throughout the days that taught me a lot about forming our pitch and focusing on what the product aimed to do even if we could not achieve total functionality. The pressure of completing the entire project in a short amount of time forced me to prioritize and split up tasks efficiently. I was happy to have chosen a project that I am so passionate about and hope to continue developing.
Future Directions
In October 2023, I decided to apply for the Kuzneski Innovation Cup through the University of Pittsburgh Big Idea Center.
The Kuzneskis have a lot of experience investing in startups and have invested around $18,000 in student teams for the last eight years. I believe that their support could help in the further development of the ML model and app.
While we did not make it to the final round, we were awarded a small prize to help get the project started. In addition, with the resources and connections the Kuzneski's have, we can conduct further research and develop a business plan for our invention.
This project is currently being developed further.