During this pandemic, everyone feels under immense stress, which significantly impacts our day to day functioning. Many students (ourselves included!) find it difficult to concentrate during lectures when sitting in front of our screens all day.

What it does

The solution uses EEG electrodes positioned in front of the frontal and temporal lobes to classify attention states. These EEG signals will be streamed to a cloud storage database system for signal processing and extraction of key features. Upon detection of an unfocused state, a timestamp is recorded and a signal is sent via bluetooth to a device called the "FocusCube". The FocusCube will vibrate and the user can press a button to indicate that they have started paying attention again. Timestamp information will also be displayed to users via a mobile app, so they know where to return to in their lecture if they stopped paying attention and so that they can track their attention performance.

How we built it

We used an SVM classifier made using the sklearn toolbox. The features extracted from the EEG signals are the spectral power at 0.5 Hz binned frequency bands from 0-18 Hz. The model parameters were saved into pickle and loaded for our real-time simulation which used a .csv file which combined data from concentrated and non-concentrated states.

Challenges we ran into

Finding datasets to test our data was an important part of our project and difficult to find. There was also some trial and error involved in building the realtime simulation.

Accomplishments that we're proud of

Coming in with little experience in machine learning, we were excited to create a classifier with 78% and see it working with demo data!

What we learned

Overall, we learned many things throughout this hackathon including; using an Support Vector Machine classifier in python, the function and importance of EEG sensors and their practical applications, how the brain functions under non-acute stress and machine learning. All of these things learned at this hackathon can be further applied to work-life and school-life.

What's next for iFocus

Some next steps include building an app and integrating the device with hardware (EEG set and FocusCube).

In the future, we plan to launch our product by:

  • funding from local universities and big tech companies (e.g. Apple)
  • social media platforms and collaboration with other companies

We also plan to further improve iFocus by optimizing the algorithm, so the device would have a wider customer demographics, including drivers and medical professionals. We hope iFocus will be applied to:

  • avoid distracted driving, or driver fatigue
  • diagnose mental illnesses at a early stage

Built With

Share this project: