Inspiration
Got the idea from a club project
What it does
Detects PD from spiral drawing
How we built it
Tensorflow model -> port to javascript -> webpage
Challenges we ran into
A lot, like figuring out how to debug a webpage (i've never done this before), converting canvas data to tensorflow tensor format, porting a python tf keras model to javascript, and hosting JSON + bin files on https for the model transfer
Accomplishments that we're proud of
Building an 87.5% accurate ML model for PD detection using static spiral tests
What we learned
What's next for Parkinson's Detection Spiral Test
Fix the model's implementation (bug that leads to printing the same value every time), implement other tests like the clock test or dynamic spiral test
Built With
- bootstrap
- html5
- javascript
- python
- tensorflow

Log in or sign up for Devpost to join the conversation.