Avatar for Graham_Jones


Member since Aug 2020 • Last active Aug 2020
  • 0 conversations

Author of OpenSeizureDetector - free, open source tools to detect epileptic seizures and raise an alarm to warn carers that someone may need assistance. http://openseizuredetector.org.uk

Most recent activity

  • in Bangle.js
    Avatar for Graham_Jones

    Thanks. If it did stop sending data when there was no movement, it would have been an issue for my automated dault detection. But it sounds like it is ok.

  • in Projects
    Avatar for Graham_Jones

    A brief update. I have modified my OpenSeizureDetector Android App to receive acceleration data over bluetooth, which is sent using the widget that @Gordon has written (thank you for doing that, and getting a watch to me so quickly so I can try it out).
    I convert the acceleration figures back into mili-g which is what my existing analysis works on.

    First results are promising - a gentle shake of the arm with the BangleJS watch on it gives a
    sensible looking acceleration spectrum, and my app generates "seizure detected" alarms as we would expect (screenshot of phone app below).
    There are a few things to add to get this to a working seizure detector:

    1. Check reliability - we need to make sure it copes ok with disconnection and reconnects automatically.
    2. Send Watch battery level every now and then (to get rid of the red 0% in the screenshot below).
    3. Send heart rate every few seconds as well as acceleration data.
    4. Have the watch receive the seizure detector status (OK / Warning / ALARM) to alert the user in case it is a false alarm.
    5. Have the watch respond to button presses to mute the alarm system if the user is doing something that he/she knows will cause a false alarm (like brushing teeth).

    Connection reliability is the most important thing to test though - will let you know how that goes...



  • in Projects
    Avatar for Graham_Jones

    some people (approx 20%) who have seizures have prewarning with signs such as aura, etc >- not sure if heart rate is one of the prewarning signs but it might be:
    perhaps an option to capture heart beat data for some period of time in a rolling log, eg. last > 5 minutes prior to seizure to compare to heart beat data immediately prior to seizure with
    aim to see if change of heart rhythm can be used to warn of oncoming seizure to allow
    sufferer to get themselves to a safe area, lay down and inform others in the area

    Yes, one of the OpenSeizureDetector users is very interested in the heart rate monitoring side - she thinks there is an increase in heart rate above the normal resting one before a seizure. Your idea of a change in rhythm is an interesting one which I hadn't thought of - would maybe need to measure heart rate variability to see if it was becoming more eratic?

    medical staff always ask how long did the seizure last; carers, at least for the first few
    seizures are unlikely to time a seizure, or there might be no-one to witness the seizure. App
    could time the convulsions.

    It is possible to extract this from the logged data, but I don't have a nice easy display of it, so that would be a good idea - maybe have the phone (or watch) display the time since the seizure started - that would be useful for people offering first aid because they should call an ambulance at 5 minutes if the seizure hasn't stopped.

    Good ideas, thanks!

  • in Projects
    Avatar for Graham_Jones

    Thank you! Remember a lot of the things in my Github were failures though, but I had an interesting time trying to make them work!

  • in Projects
    Avatar for Graham_Jones

    This sounds like a great idea, thank you for getting in touch, Gordon.

    I think the documentation in the Github repository is a bit sparse - most description of OpenSeizureDetector is at http://openseizuredetector.org.uk - look at the Garmin Seizure Detector part to see what it does.

    If I modify the Android App to receive BLE data directly we should be able to make it work with other devices, which would be great - I'll look into how to do that now.
    If anyone is interested what we have looked at for seizure detection, the slides of a talk I gave today are now here: https://github.com/OpenSeizureDetector/P­resentations