Seizure detector #1800
Spiralo-Idioide
started this conversation in
Ideas
Replies: 2 comments
-
This may help. Not sure what the progress is with Pinetime. https://www.openseizuredetector.org.uk/?page_id=1480#PineTime |
Beta Was this translation helpful? Give feedback.
0 replies
-
Excellent idea. This is very close to what we have with OpenSeizureDetector. There are currently two 'nearly working ' implementations for pinetime (we do all the processing on the phone so the watch app can be quite simple). If you have a look here: https://github.com/orgs/OpenSeizureDetector/discussions/70, I can talk you through what would be involved in getting it working. |
Beta Was this translation helpful? Give feedback.
0 replies
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
-
Hello everyone !
The main idea
I have quite frequent seizures and my family and friends are anxious that I might have one alone or somewhere where the people around me don't know how to react.
So I had the idea to use the pinetime as a seizure detector that connects to an android phone in order to send a message when I might have a seizure.
The kind of seizures that might be possible to detect would be :
I wanted to share with the pinetime community this idea to know if it would be possible to implement.
Also I am sorry if my english is not perfect, if anything is hard to understand, don't hesitate to ask me, I will try to explain it better.
Also I mostly say "I" but it should be understand as "any owner that uses this project"
Why the pinetime ?
Possible functions
Main functions
Possible other functions
Everything here should be possible to activate/deactivate directly with tasker so it is not needed to program it into the pinetime
How I think it might work
Detector
The detector would mainly work using accelerometer data.
Heart rate data could be useful but would most likely be hard to interpret without a lot of seizures data.
Every 1 to 5s an analysis is done on the data to detect a possible seizure.
Accelerometer
Help/ideas are needed here
Last 30 to 60s of accelerometer data would be stored in the ram with interval of 0.05 to 0.5s.
The analysis would most likely try to interpret the acceleration as movement on the axis. If a lot of fast movements are detected on the 3 axis at the same time, then the seizure detector considers a seizure is detected.
At least some data will be needed to have as little false positives/negatives.
Storage
Help/ideas are needed here
Storage will work with files stored every 5 to 60mn. The name of the file will be the time where the file data begins (rename it if a seizure is detected during the time range it covers).
If possible the best would be to store the data on the user phone instead of on the watch.
File size for accelerometer
I am not sure how to store raw accelerometer data without having file sizes impossible to store in the watch memory which is very limited.
Maybe it might be possible to lossy compress the data on 2 to 4 bytes per axis and not storing the data every 0.05 second.
Min file size :
2 bytes per axis, stored every 0.5s.
(3 x 2) x 7200 = 43.2 kB or 0.0432 MB per hour
1,0368MB to store the last 24h.
Max file size :
4 bytes per axis stored every 0.05s.
(3 x 4) x 72000 = 864 kB or 0.864 MB per hour (definitely impossible to store on the watch)
File size for heart rate
Because heart rate data doesn't need high precision, storing the average of the last 5 to 10s will not be a problem at all. It is even possible to store as 2 bytes, HR over 255 being very rare.
File size would be 0.72kB to 1.44kB per hour.
17,28kB to 34,56kB to store the last 24h.
Conclusion
I have no experience with the pinetime (I don't even own one at time of writing) and because of that it is possible that 95% of my ideas are materially impossible. Maybe the core idea in itself is entirely impossible to achieve.
Anyway, thank you to everyone that will read this post. And thank you to everyone that will take part on the discussion, even if it is to explain why this project is impossible to achieve.
And thanks again !
Beta Was this translation helpful? Give feedback.
All reactions