My app uses the Muse headband to collect EEG data from users. The data has 4 channels that are along the frontal lobe. I gathered 1000s of EEG data points from multiple people while they were in focused and non-focused environments. I used this data to train a machine learning model to calculate the focus of a user on a 0-100 scale. The Pytorch neural network (coded in Python) was able to cut through the noise and identify how to measure focus due to the large dataset and use of multiple people; it was able to generalize. The app splits the raw EEG into 5 frequency bands by using FFT. It then feeds the gamma and beta waves to the AI model. The app plays music from a playlist of downloaded music and assigned a focus score to each song. It then lets the user know which songs are helping them focus along with insights on which genres help them focus. They can see their real time focus and music-focus data in a web app. The app is built using a Python backend in Flask. It was designed in Sketch and uses HTML for the front-end. The app is really useful for students because we study for a living and most of us listen to music while doing it. The problem is research hasn’t clearly shown whether or not music helps your focus and everyone’s brain is different so the answer varies for people. Intelom customizes to everyone.

 

What inspired you (or your team)?

My inspiration was when I began researching brain-computer interfaces. I learned how EEG could give us brainwave data in an accessible and non-invasive way. I had also taught myself machine learning on the internet and thought this was a great application of AI. The two are perfectly paired technologies. One is the science of data collection, the other of data processing. I love listening to music while studying but certain songs help me focus while others distract me. Research papers haven’t reached a consensus so I decided to create my own solution. The hypothesis that inspired my project is that we haven’t reached a consensus because everyone’s brain chemistry and neutron wiring is different and since we all know different songs and have different music preferences, music’s effect on focus must be different for everyone. I love my project and now use it often while studying to create some great focus playlists that are constantly updating based on my brain! With studying being one of the most important things we do in life. Just completing high-school means you’ve spent around 19,000 hours studying. Because of this, I feel that a tool that helps you choose better music (or tell you that all music hurts your focus) is an extremely valuable tool for students around the world, or really just anyone who’s trying to get work done while listening to music. Also, Please note that the project video was created during prototyping and does not demonstrate the new user interface. Also the focus scores in the video are on a 255 scale, while the updated app shown in the project image uses a 100 scale.