Focus brain data tracking
Focus prototype dashboard
Focus real time interface adaptability
Ever since the first time we met together as a group to prepare for this hackathon, the recurring theme in all our brainstorming sessions was that we wanted to help students learn better. Each of us has had their own struggles in the education system and we all agreed that this was something we were passionate about. The question quickly became: “How can we create a more efficient way for people to learn?” As we clarified our passions, we quickly saw that we were interested in using emerging technology like brain wave tracking to solve existing problems in innovative ways. From the merging of these two areas, the idea of FOCUS was born.
What it does
FOCUS is designed to maximize the effectiveness of information comprehension and retention. First, we train the model to figure out your personal trends. Then we use the brain wave tracking data (obtained via headset) to create a conversation between the user’s mind and FOCUS. Based on what FOCUS receives, FOCUS will adapt the web interface to display information in a manner more suited to how the user is feeling in real-time. This same data will also be stored and displayed in a dashboard interface where the user will be able to see their attention growth over time. Characteristics of their most and least productive hours will be displayed so that user will be able to adapt their own habits based on what they want to achieve. All of this functionality will come together to provide a convenient and high impact way for a user to make the most of their time while learning.
How we built it
Once we successfully configured the device to send data via Bluetooth, we obtained the alpha and beta wave ratio of the user. From there, we used Tensor Flow to apply machine learning on the data using classifiers. After, we extracted information about the focus levels at a certain accuracy and sent this data to the front-end.
On the front-end side, we built a chrome extension that manipulates the adaptability of the interface. It does this by using a web socket that receives data from the backend. As additional data is received, the interface will continue to streamline itself.
Challenges we ran into
There was definitely a learning curve for Tensor Flow and Google Cloud Machine Learning. Specifically, understanding how hypertuning and creating a model works. We also spent a lot of time obtaining neural data from the NeuralSky device. Figuring out how to send data via Bluetooth serial port to the server from the device was a challenge as well.
Making sense of the brain wave data is still one of the challenges we are facing. After analysis, we realized that beta to alpha ratio makes sense to track because it yields beneficial insights into the user’s focus level. On the front-end side, we also had to figure out how to modify the interface in real time through chrome extensions.
Accomplishments that we’re proud of
We wanted to push ourselves to learn about technologies we didn’t have much exposure to in the past. As a team, we’re proud that we were able to learn this much about Machine Learning in such a short amount of time. Not only that, but we were also able to make a lot of progress on modifying site structure in real time. On the user experience end, usage of these rapid user interviews and multiple group brainstorming sessions to synthesize our findings allowed us to create a solution we are proud of.
What we learned
We learned a lot about Machine Learning, Tensor Flow, integrating with NeuralSky, and modifying site structure in real time.
What’s next for FOCUS
Explore more into the best metric to measure in order to understand more about user focus levels. We want to take this product to a place where we’re able to correlate characteristics of their external states with high and low focus levels. We would then be able to use this to create real time recommendations that are backed by data.