Health Diagnoser

Ever feel pain in your body, but don't know what it's called, or if you should do something about it? The Body Diagnoser solves exactly that - know why it hurts!

Comment

Inspiration

We were inspired to create the Body Diagnoser after seeing the third SDG and its ability to help people in desperate need. Expensive out-of-pocket payments have pushed 381 million people into extreme poverty, and we saw an opportunity to help millions of people through a free website for all.

What it does

The Health Diagnoser is a public website that anyone can use to quickly self-diagnose themselves for free. It asks for pain point and a description of the pain

The Health Diagnoser is a web app that uses a specially trained AI to diagnose health issues based on the user's description of pain and other symptoms. The user is asked to select points on their body where they feel pain, and then describe it and any other symptoms they are experiencing. The diagnoser will then produce a diagnosis for the user and suggestions on what to do.

How we built it

We designed the idea and web interface on Excalidraw. We wrote the code, with some help from AI, in PyCharm and VS Code and did some debugging to fix edge-case errors.

Along the way, we pushed some commits to our GitHub repo.

Challenges we ran into

One of the biggest challenges was defining (and naming!) regions across the entire human body, to allow the user to be specific enough in their selections. Doing this correctly also tells the AI where the issue is, so it was critical to take our time and get it right.

Another challenge we had was getting the model to even render on the page - 3D model rendering and click-handling is surprisingly difficult! Eventually, we got it working with Three.js.

Oh, and that one time we couldn't get the code to work because we forgot to hit save... 😅

Accomplishments that we're proud of

We're proud of beating all of the challenges listed above. Apart from that, we're proud of correctly fine-tuning the AI to get us an accurate diagnosis most of the time.

What we learned

From this project, we learned a lot about Three.js and rendering, along with training AI to produce the results we need. I (Lakshya) learned a lot about how OpenRouter, Flask, and PythonAnywhere work, and I (Bhuvan) learned how to render 3D models!

What's next for Health Diagnoser

We're hoping to, in the future, have even more accurate diagnoses and better suggestions for what to do next. Additionally, we could implement a feature that allows users to download/print their results. Minor updates like these are what we're thinking of doing to improve Health Diagnoser.

Built With

Share this project:

Updates