fairHealth | Devpost

Inspiration

We have experienced firsthand, through our family and community, the lack of transparency in healthcare and the desire for accessible medical information.

What it does

This app analyzes hospital bills to detect potential overcharges, providing transparency and financial clarity to patients. Additionally, it uses symptom inputs to deliver accurate medical diagnoses, offering accessible healthcare information to users.

How we built it

This app is called fairHealth, built with a custom Generative Adversarial network, hand-made location API, Recurring Neural Network, and more technologies. It is very highly technical. There are model accuracy statistics. The main 2 features for the app are: 1. A feature that analyzes your medical bill based on your location to give you an accurate estimate of how much to be charged (overcharged or not + actual cost). According to the ABA, hospitals can charge unto 10X more than actually required, and many are scammed under their innocence. We used a local server.js file and mongoDB. The medical bill is processed/parsed and then goes through our GAN while locationAPI is called and then cost is calculated and individual items are compared with public access/ acquired datasets from NIH, CDC, and Zuckerberg County Hospital. 2. Symptom diagnosis with another AI model. This is designed to reinforce the concept of being independent from hospitals/wasting money. Very accurate to avoid Google-type of over exaggeration. Detail and focus on the 1st feature and throw in many visuals.

We used MongoDB, a no-SQL database integrating a local server js on port 5175 etc., and blockchain/encryption technology to preserve integrity of user credentials as well as medical info. emphasis on this

We used Python and PyTorch for the GAN and the RNN. The backend was done through React Native. Front end was done through HTML, CSS, SCSS, JavaScript. Everything was built from scratch.