Inspiration
Our inspiration is the role of data in health-care. We truly believe with more data points, and a comprehensive analysis of these data points, individuals can truly understand their mental and physical well-being and make informed decisions about their health. We really think the cost and sometimes the lack of quick and affordable mental-health diagnostics can serve as a barrier to helping people get the care that they deserve.
What it does
Our product leverages a powerful model built on an extensive Quantitative EEG dataset from more than 850 patients to accurately assess mental health disorders. By analyzing raw EEG data and then transforming this into Quantitative EEG data, our system can predict whether a person is likely to be suffering from Addictive, Anxiety, Mood, Obsessive Compulsive, Schizophrenia and Stress-related disorders. Our initial feature is a program that transforms an individual's Raw EEG Data into a Quantitative EEG format that our machine learning model can use to predict an individual's mental health disorder. This program utilizes the Butterworth low-pass filter in Scipy's Python Library to attribute raw EEG data to different bands (delta, theta, gamma, alpha, beta and high beta) that are used in data processing. Our machine learning model is a proprietary model that leverages Support Vector Classification to accurately classify an individual's mental health disorder. This model was implemented with a train-test split of 0.3, and was optimized with hyperparameter tuning. When combined, individuals can update their raw EEG data to our website and they can get the mental health condition that they are likely to experience (or if the data is healthy, than a health condition will be provided). Additionally individuals can be directed to specific resources on their diagnosed mental-health condition.
How we built it
The program to transform data and the machine learning model was built entirely in Python. From there we utilized Flask to combine these features with our website which was made in JS, HTML and CSS.
Challenges we ran into
The biggest challenge was converting raw EEG data into quantitative EEG data. To do this, we had to learn about signal processing, and ultimately we applied a Butterworth signal and zero-phase filtering combined with Fourier transformations to convert the data.
Accomplishments that we're proud of
This is the first hackathon for all of us, so we're super proud to participate and bring to life something that was just and idea 24 hours ago.
What we learned
We learned how to work in a team, how to take an idea and turn it into a product, and the importance of time management.
What's Next for Simplify
We want to automatically update our ML model when users input their own data and later on, we hear from them after multiple follow-ups. This was we can scale our ML model and ensure the accuracy improves over time. Next we would love to finish the website, and deploy it so everyone can use it. We think our product is really niche and it can be very useful in the fields of Neurology and Clinical Psychology.
Log in or sign up for Devpost to join the conversation.