Skin.AI

Inspiration

Cancer is a burden that has affected society throughout the centuries and to this day a cure has not been created but with the advent of machine learning, Cancer can be detected in its early stages and possibly save the lives of people around the world.

What it does

SkinAI uses machine learning to analyze a photo of a mole a person may find concerning and gives a prediction of the likelihood of it being benign or malignant with up to 86% accuracy. The app will also give daily fun facts for the user to help with skin cancer prevention. It is also on the network edge giving the user the benefit of not needing an internet connection, low latency, and having more privacy.

How we built it

We took a service-oriented approach. Apple CoreML holds the neural network for detection. It communicates with the Client iOS app, where business logic and predictions are handled.

Challenges we ran into

Getting a reliable and large enough data set to train the machine learning program. The program was reading the malignant and benign backward. We also ran into an overfitting problem.