MelaModel

LandingScreen
DoctorScreen
LoginScreen
UploadScreen
ResultsScreen

Track: Access to Healthcare

Inspiration

As outdoor sports players in Southern California, we are always worried about our skin health. The UV radiation from the sun always leads us to have sunburns and unusual mole growth, even when we apply sun-screen. While we are dealing mentally with the game, we also deal with our parents who ask us "what is this growing on your skin" or "did you put your sunscreen on". We would always tell our parents that this mole is fine, however one day one of our players mole developed into melanoma. While we noticed this unusual growth early, we were wondering about the other people who were unaware of melanoma or simply did not have the resources to check for melanoma. People might not be able to differentiate between a harmless mole and a dangerous cancer, and hospitals are overwhelmed with COVID-19 patients. This is where we developed MelaModel, an app that detects for melanoma through your smartphone.

What it does

MelaModel is a simple but powerful application built into your smartphone that detects the presence of melanoma in a photo of the skin using Artificial Intelligence.

How we built it

We built this iOS application using react-native, while using Tensorflow to train our models. Our backend was deployed through Amazon Web Services.

Challenges we ran into

A challenge that we ran into was assigning jobs to each team member. While we all specialized in one specific aspect of the application, integrating them together into one app was difficult because sometimes we would need to wait for one teammate to finish their part to continue on the other part. Although this was problematic, by building each individual puzzle piece together, the process became more exciting.

Accomplishments that we're proud of

An accomplishment that we were proud of was completing the application in-time. As this application required a lot of teamwork and dedication, each team member was able to work in a timely-manner and in an organized manner. (If you check our group chat, we really pushed each member to there hardest ensuring that the final project is successful)

What we learned

One thing that we learned the most was the medical concept behind melanoma. At first, we only knew that it was a serious type of skin cancer, rather now we now the different signs and indicators that can lead to a strong indication of melanoma. For example, if one of your mole is not symmetrical, then that could be a strong sign of melanoma. Honestly, this experience allowed us to have fun, while also learn more about medical and computational concepts. (At the end of this project, all of us took random pictures of any skin mole that we had and tried plugging into the AI to detect if it was melanoma: what a great way to end this project for all of us).

What's next for MelaModel

One thing that we want to have next for this iOS application is by having allowing patients to view all their past scans. This feature would be helpful because patients and doctors would be able to view the development of their mole, specifically determining if the mole is growing into a more serious conditions that would require immediate medical action.