mediCAM

Inspiration

One of our team members is passionate about medicine and applying technology to it, so she suggested creating a portable healthcare app. Together, we brainstormed how to create this and decided to use visual recognition and neural networks to achieve our goal.

What it does

mediCAM takes a picture of physical, external symptoms and uses machine learning to classify it by percent likelihood of the diagnosis. Based on these possible diagnoses, the user can learn about the diagnosis, the best course of action, and other symptoms.

How I built it

We used python and tensorflow to create the machine learning software. We then downloaded a database of melanoma (skin cancer) images for the program to classify into benign or malignant. We used Sketch to outline our UI, then transferred our plans to Android Studio.

Challenges I ran into

We spent a lot of time trying to use a python-based app developer, before realizing that we didn't need it.

Accomplishments that I'm proud of

Our program was able to train itself and we learnt a lot along the way.

What I learned

What's next for mediCAM

Including a survey so users can input internal symptoms and narrow down the possible diagnoses.

Built With

• android-studio
• keras
• matplotlib
• numpy
• opencv
• python
• sketch
• tensorflow