Epidemiology is critical in figuring out how to stop the spread of diseases before it’s too late

What it does

ClassiFly uses image data to classify individuals with known disease symptoms. For demonstration purposes, we selected Yellow Fever and Methicillin-resistant Staphylococcus aureus, and Eelephantiasis.

How I built it

The app was developed in Swift, and the classification model was trained using a split data classifier method, which leveraged Apple's native CreateMLUI framework to build an image classifier model with 89% accuracy.

Challenges We ran into

We initially planned on building an autonomous drone for tracking that could be used to identifying certain key epidemiological characteristics in a medically unsafe and infected region. That is, this would effectively increase accessibility to remote areas that are susceptible to infection. However, there was no clear way to interface with the drone via an API so we decided to simply build a classification app that would allow you to use a drone image taken in a contaminated area and derive certain key epidemiological insights.

Accomplishments that I'm proud of

I am proud that we were able to successfully work together efficiently to build an image classification app.

What I learned

We learned how we navigate managing development projects as a team, as well as how to leverage really powerful computer vision capabilities with CoreML.

What's next for ClassiFly

What if we could have an army of medical detectives in the sky, able to reach the most remote populations? Briefly: Navigate to remote areas, collect image data of populus, use Machine Learning to classify afflictions based on visible symptoms. This paints a better picture of the disease landscape much faster than any human observation.

Built With

Share this project: