Keisaku

Training a Medit-AI-tion assisitant

YouTube Video

Step 1 - Collecting EEG data from an OpenBCI headset for later processing

Out of conveninence/laziness, I have chosen to use the OpenBCI GUI rather than configure a direct USB connection and use BrainFlow natively. Therefore, this repo implements a simple JavaScript UDP server which streams the received data straight to a CSV.

Assumptions / Usage

1. Start the Keisaku App by running node udpServers.js
2. Start up the OpenBCI GUI and configure the Networking Widget to stream UDP to the Address running Keisaku. In my case, I am streaming from Windows to Windows Subsystem for Linux. The IP is printed by the Keisaku once it is run.
3. After ensuring the headset is collecting good brain data (i.e. running impedence checks, etc), click Start UDP Stream
4. In the keisaku terminal, press s to being collecting and storing the streamed data. The App is currently setup to collect and store Aux/Digital Button Press, Focus Widget Output, and the Raw Timeseries data (with filters). There are additional methods for collecting & parsing BandPower (currently commented out). Any other GUI output can be enabled by creating a parsing function (based on data format as definied in the OpenBCI Docs).
5. When you're done meditating, click Stop Data Stream and System Control Panel > Stop Session. This allows the GUI to save its raw Session data (which Keisaku will look for and try to upload). Then press c in the console to close the UDP servers and begin uploading the data.

Step 2 - Uploading Collected Data to a Cloud Datastore

6. My repo is set up to upload the data created by the UDP servers, as well as the latest GUI-generated raw data, to an S3 bucket defined in a .env file in the root of the repo. Store your corresponding AWS credentials (AWS_REGION, AWS_KEISAKU_ACCESS_KEY_ID, AWS_KEISAKU_SECRET_ACCESS_KEY) in a .env or modify for your Cloud Store of choice.

Step 3 - Work with the Data in your new Data Lake using a Data Platform of your choice.

7. I've personally chosen DataBricks to mount S3 Bucket and begin exploring my data. However I'm not an expert in the whole platform ecosystem. There may be a better option for price and ease-of-use considerations. Or, of course, you can start writing Jupyter notebooks locally to explore your data.

Step 4 - 🧘🏻‍♂️

8. Collect more data.
9. Stay tuned for updates here. And share with me your updates!