SkinCheck Bot

SkinCheck Bot analyzes uploaded skin lesion images to predict cancer risk using a trained AI model, and provides educational responses to user questions by referencing real clinical skin cancer data

Comment

Inspiration

Skin cancer is one of the most visible and common forms of cancer. We wanted to explore how technology can help with early detection, while also providing information about the differences between malignant and benign skin lesions.

What it does

Our web app, SkinCheck bot, allows users to upload images of moles or skin lesions and uses a machine learning model to predict whether the mole is likely benign or malignant. It's built to be simple and accessible for anyone to use.

How we built it

We used a skin cancer image dataset from Kaggle and trained a model in Google Colab. Then we connected it to a Gradio web app so people can upload their own images and get a result. We also used Docker and MINDS to pull in medical data.

Challenges we ran into

We ran into a few challenges while working on this project. Setting up Docker and the PostgreSQL database was confusing at first since it was our first time using them. Some of us were also new to Google Colab, so getting used to it took some time. We had issues with the runtime disconnecting, which slowed things down. Hosting the app with Gradio didn’t work right away either. On top of that, we had to fix problems with file paths, organizing the data, and making sure the image predictions were accurate.

Accomplishments that we're proud of

We started with a simple idea and turned it into a real working web app. Our model can predict if a mole is benign or malignant based on an image upload. For some of us, it was our first time using tools like Google Colab, Docker, and Gradio, but we were able to learn and figure it out as a team. Getting everything to run how it's supposed to was a big win for us.

What we learned

We learned how to train and test a machine learning model using images. We also figured out how to use Google Colab, Gradio, and Docker to build and run our project.

What's next for SkinCheck Bot

We want to make our model better by using more pictures so it can tell the difference between benign and malignant moles more accurately. We also want to turn SkinCheck Bot into a mobile app. That way, people can take a picture of their mole with their phone and get a quick answer.

Built With

Share this project:

Updates