Inspiration
The American Medical Journal reported 100 000 annual deaths due to misdiagnosis every year.
What it does
Pneuva aims to provide a sanity check for doctors by comparing an uploaded chest x-ray against 10 000 verified data points.
How we built it
Using Microsoft Azure Custom Vision AI we trained a custom model using a data set found on Kaggle. We used html with javascript and CSS for the front end.
Challenges we ran into
We originally wanted to use google cloud vision but were unable to format our data points correctly and had to pivot at the last minute to use Microsoft Azure. We also had some trouble getting the API keys to work with our front end.
Accomplishments that we are proud of
Our 2 person team started this hack around lunchtime on Saturday and we were able to finish!
What we learned
Structuring your data is very important when training an algorithm.
What's next for Pneuva
We hope to onboard doctors and hospitals to grow our dataset of images as well as provide more information
Log in or sign up for Devpost to join the conversation.