Inspiration
Diagnosing a psychological disorder can be more difficult than you might expect. According to a study by Zeigler et. al. from the University of Leipzeg, for disorders like OCD diagnosis after the first onset of symptoms can take up to 17 years, and the mean amount of time before a diagnosis is 12.78 years.
Our product, NeuroScan, promises to change that. By using a neural network to process EEG scans, psychiatrists can get valuable information about their patients' conditions. While psychiatry is generally an under-represented field in terms of AI, our product aims to change that and revolutionize the industry
What it does
Our product predicts patients' psychological conditions using EEG scans!
How we built it
We built the front-end of our website using React, and the backend using SQL and Amazon buckets. The neural network was built using python, PyTorch The whole thing is hosted on an AWS server.
Challenges we ran into
Our model initially didn't perform too well. We were forced to change it twice, eventually settling on an architecture that doubled our validation accuracy! We use a combination of different models and choose the best one possible for each disorder.
Accomplishments that we're proud of
We have a full working model and functional front-end and back-end hosted on a server. It went better than we expected, the team was in sync and worked very well together
What we learned
We learned how to use pytorch and many different models, how to use React, and how to use AWS buckets.
What's next for NeuroScan
We plan to keep working on the model, we're first need to work on increasing the accuracy and finding more training data. Once we do that, we plan on finalizing the frontend and marketing the app to psychiatrists!
Log in or sign up for Devpost to join the conversation.