Inspiration
The night before when we were talking about an idea for the hackathon the next day, we realized that affordable healthcare accessible isn't available in the U.S. which is why we decided to make a custom model of GPT-4 for health diagnoses.
What it does
Our custom-built GPT-4 model returns a diagnosis of a potential disease you may have based on your symptoms and looks through our database of over 20,000 drugs currently being sold in the United States to return drugs needed to help with treatment and also their cost also factoring in the average number of tablets/bottles per pack.
How we built it
We built this web app using Python and React making use of the GPT-4 Turbo assistant retrieval. Significant prompt engineering and testing along with making use of React to code the frontend is the bulk of this project.
Challenges we ran into
Some challenges we ran into were limiting the number of words per response in order to make sure we don't hit rate limits from our assistant. We also were not able to implement a Python API to integrate the frontend with the backend.
Accomplishments that we're proud of
We are proud that we were able to do very good prompt engineering and python backend development as well as a good amount of last minute frontend development.
What we learned
We learned that app development is a process that requires good frontend and backend portions of the project and also time management is important.
What's next for Healio
Our next steps will be to soon implement a Python API to connect the frontend and backend while also developing some of the features we wanted to include but were only able to implement the hardcode for. This includes the past conversations section which saves the message threads from Healio and also the nearby doctors function for quick consultations nearby, ideally forwarding the message thread to the doctor so that the doctor is informed of the patient's condition and can take action as early as possible.
Results
3rd place in Health Track
Log in or sign up for Devpost to join the conversation.