1 in 20 people is misdiagnosed each year in the United States based on the data from medicalnewstoday. That is equivalent to 12 million people. Misdiagnosis can be threatful to human life. We want to change that.
What it does
Doc.AI is a tool for doctors that will assist doctors in diagnosing various diseases by scanning X-Rays and MRI scans with an iPhone and/or iPad. We are starting with four diseases: Brain Tumors, Pneumonia, Tuberculosis, and Scoliosis
Due to the simple UI, Doc.AI is easy to use:
- Choose the disease
- Point the camera at the X-Ray or an MRI scan
- Tap the shutter button
- Get the prediction
That is not all!
World-known doctors will be able to become a Doc.AI contributor.
We want to help doctors’ better diagnose and provide proper treatment for their patients, and to do that we need doctors’ to help us diagnose those patients.
Through a leaderboard-led scoring system, qualified doctors can help train our models by answering simple questions about our prediction, in case it was incorrect. They will also be able to compete against each other by earning DocPoints each time their contribution is accurate.
Lastly, once Doc.AI gives a prediction, the doctor can send the prediction information to the patient's primary physician for further treatment
How we built it
The app was built using Swift and CoreML framework to classify images. In addition, we have trained four models (for Brain Tumors, Pneumonia, Tuberculosis, and Scoliosis).
Challenges we ran into
Training your own model sounds easier than it is. This process is very time-consuming. We had FOUR. Due to this event being only 24-hours, we had to work in a hurry to get as much training data as possible. The rest of the time was spent on making the UI pretty and testing the models.
Accomplishments that we're proud of
- Trained not one, but four models
- Made a sick UI
- Squished most of the bugs
What we learned
The biggest thing we learned was teamwork. This project would not be a reality if we would not have good teamwork and composure.
The team worked non-stop and each member was participating in every area of the project
What's next for Doc.AI
- Improve model accuracy
- Improve the security of the app