Many people struggle to pay exorbitant bills, including simple consultations they may not need. Using machine learning, I built a model that can pre-screen an individual's skin lesions!
Inspiration: My father had to get a skin lesion inspected, and it cost an exorbitant ammount of money. I soon began to realize many others suffered from this issue, catalyzing my search for a solution. I want to make simple pre-screening in American health care accessible!
What it does: Using machine learning, this model deciphers whether your skin lesion is a BCC, SCC, keratosis, benign, or cancerous!
How we built it: We used google colab software, python, and sklearn!
Challenges we ran into: Debugging was difficult, and had to install a plethora of dependencies.
Accomplishments that we're proud of: Enduring the debugging process!
What we learned: How to build a reliable model!
What's next for Skin Lesion Image Classifier (Using Machine Learning): Definitely increasing accuracy with more data
Judges: Pigmented Keratosis falls under benign and vice versa! Same with cancerous melanoma with BCC and SCC, impacting my accuracy; They are however medically interchangeable. If you submit a PK and it comes back benign, that is a correct prediction. Definitely something to improve upon in the future of my model, perhaps by lessening categories to simply malignant and benign.
Log in or sign up for Devpost to join the conversation.