Inspiration
As doctors in the health industry, my cousins are always complaining about how difficult using the computers while at work are. "They're too old! They don't give specific enough data! Sometimes they're wrong!" Among the clinical technology hardware innovations booming across the planet such as a mantis shrimp's hexachromatic bioinsired camera lens that's able to more find and identify tumors, hides the incredible software innovations such as AI. I wanted to make something that was able to make a doctor more efficient, and confident in their results.
What it does
Identify whether or not a tumor is benign or malignant based on an array of different factors.
How I built it
Google's CoLab
Challenges I ran into
My program simply could not run-and it still can't. It ran at one point last night where I got a success rate that hovered around the 93-97% accuracy for the 20% (113 data points) of samples I saved from the dataset to use for the testing after I trained my AI.
Accomplishments that I'm proud of
Incorporating all the steps of making an AI, doing research, making something that will make a difference in the world.
What I learned
data segmentation, the difference between data testing and training, utilizing the different frameworks offered by CoLab, patience, becoming a YouTube and Google Search by using keywords.
What's next for Breast Cancer AI
Getting it to run successfully and it being more dynamic so it can take in more data--expand the AI to identify more factors for other cancers.
Log in or sign up for Devpost to join the conversation.