Pebble Seizure Detect | Hacker News Reader

Pebble Seizure Detect

github.com

301 pointsingve46 comments

Always warms my heart to see technology re-used in interesting ways that can help people with a medical condition.

Way back(~2004) I was working for UPS. One of our package loaders had a BlackBerry, back when only suits had them. He was mute and since he couldn't speak he used the BB for all of his communication. Here I was just geeking out over smartphone tech(Kyocera 7135, woo!) and he was able to have his whole life changed in a fundamental way with tech that was conceived for business productivity.

I once wrote an article about how the Danger's Hiptop phone (aka T-Mobile Sidekick) was revolutionary for the deaf. Pretty sure it was the first cell phone that interfaced directly with the TTY network from the phone keypad. It was a pretty neat little device actually. Ahead of its time in many ways.

I used to work at Subway ~10 years ago and a deaf couple would always come in and order with their sidekicks. They would even tell us jokes! Thanks for digging up that memory for me :)

Very much so, it was the first consumer device that had push-email, Danger was very much ahead of their time. You can see lots of things from it that influenced Android when Andy Rubin went over to Google.

Absolutely great; thanks 'ingve for posting it.

One thing that IMO should be modified is the dependency on Twilio. I don't know how it is in the US, but Android users in Europe should not need an Internet connection for sending text messages - the phone is capable of doing it by itself. I think it's more common to be without Internet access but with mobile service than the other way around.

EDIT: submitted the suggestion as a GitHub issue.

> I don't know how it is in the US, but Android users in Europe should not need an Internet connection for sending text messages - the phone is capable of doing it by itself.

On iOS, I'm fairly certain it's not possible for an app to send a text message from the phone by itself, without user intervention. This is probably a good thing; Candy Crush can't text me about my wife's latest achievements, etc.

So it's probably more of an operating system limitation than a US vs Europe thing.

You may be right. I've checked how Pebble's ability to reply to text messages works on iOS; apparently, it generally doesn't - they've introduced a limited ability for some users only recently[0]. Since the author (and/or his wife) uses iPhone, I can see now why the Twilio is the default.

Replying to text messages through Pebble works on Android though (and I've also been using Tasker to send regular texts for some time), so Android owners could use this feature. I guess it's time to write that patch :).

EDIT: The author confirmed that it is indeed an iOS issue.

[0] - https://blog.getpebble.com/2015/11/23/iosreply/

> This is probably a good thing; Candy Crush can't text me about my wife's latest achievements, etc.

Software should be allowed to help you. If you want it to be able to send a message on your behalf, it should be able to. The ability to toggle permissions per-app is one of the strengths of iOS and the limited nature of the OS is one of its greatest weaknesses.

Get 200 extra gold jelly beans by enabling SMS sharing! Just click Allow when the popup asks for permission!

That's why App Review exists – Apple could theoretically only give this ability to select apps that apply for it. They do this currently with "Network Extensions", for the ability to write a custom VPN plugin because it involves giving a third party app access to all network traffic from the device. There are always good compromises to solve these problems when you maintain 100% control over your software.

God if i was the author i'd just say "Patches/Pushes welcome."

He made it a way that works for his needs -- if you think it should work otherwise; feel free to contribute. Suggestions are a dime a dozen 'round these parts.

Suggestions without patches are extremely useful when someone gives you a use case that may be common but which you wouldn't have thought about because it's outside your own personal work flow.

Software I write for customers always has a "Suggest a Feature" button somewhere on the primary nav.

Some of the stuff suggested is ridiculous or impossible but sometimes they suggest stuff that in hindsight would have been blindingly obvious but I'm not the end user no matter how hard I try to put myself in their shoes.

I asked the author in the issue if he's planning to continue development and if he's willing to accept pull requests. I learned about this watchapp an hour ago; I didn't even have time to test it on my Pebble. I just realized that this would be an obviously important feature if someone in a similar situation to mine had to use this app.

If the author is interested in pull requests and doesn't have time, I do actually consider sitting down and writing that feature - that's the reason I asked.

There's nothing wrong with making suggestions. The author doesn't have to implement them. Being demanding is bad, but merely making a suggestion is perfectly fine.

My partner died 3 months ago, presumably due to complications of a tonic-clonic seizure. She was in the room next to me with a door closed because she was feeling unwell with a headache for the two days prior and so was dozing in bed. I was either doing the vacuuming or playing a video game at the time. This technology may have allowed me to save her as I would've seen/felt the text message and been able to provide first aid to her immediately.

For those of us who have to live with a loved one who suffers seizures, thank you to the author for making this and thank you ingve for sharing it. Hopefully it can avert a tragedy like mine for someone else in the future.

One thing I really liked here was the run through of failure modes in the readme, its refreshing to see the limitations of some software listed up front rather than glossed over.

As a father of a child with epilepsy* and who has observed many tonic-clonic seizures, I'm horrified by the thought of being an independent adult who would have to deal with them on a regular basis. So I can't upvote this enough; I feel as if epilepsy is one of the red-headed stepchildren in the age of AIDS awareness and pink-colored campaigns everywhere.

Kudos. This is awesome.

* More specifically, Myoclonic Astatic Epilepsy (MAE), which he appears to have been lucky enough to have outgrown, though I don't know if saying he "had" epilepsy (past tense) is accurate (or even relevant to the discussion).

Although my epilepsy is well-controlled, a breakthrough seizure is still a possibility. An app like this could really help me out of a tough spot. Cheers to you for building it!

For many patients, it's important to record how long their seizures last - to escalate/911 if needed, and also after the fact to aid diagnosis and treatment. I could see this app playing a role there too. After the event, the Pebble might prompt the wearer/responder whether to log it as a seizure, which would (mostly) automate the seizure log recommended by Epilepsy Society.

It's interesting to me that I've now seen two projects for Pebble built around this concept.[1] I think this is in part due to the approachability of developing on Pebble's platform. They've got a really great cloud IDE [2]

[1] now on iPhone too http://neutun.com/ [2] https://cloudpebble.net/

Cloud IDEs are a great idea for embedded devices. A friend recently showed me the IDE for Particle (formerly Spark) [0]. It definitely beats having to figure out which compiler to use, install drivers, get the right connector and get everything working together.

[0] https://www.particle.io/prototype#particle-build

Congratulations!!! It can't see a better use of technology and open source than this one. I'm really touched.

Hi all! I am the author of this software. I didn't submit this to Hacker News -- someone else must have! Thank you, whoever you are. I'll try to answer the other questions in the thread now.

This is an awesome project. Thanks for making it. My small suggestion (though it's a big task) would be to allow anonymous data collection when there's a seizure. Getting the sensor data on real seizures I am sure would help a lot in improving detection.

That is a very good suggestion! Would be happy to accept a patch if you'd like to code it :D

This is pretty cool, good job!

    // A sqrt function I got from the web that 
    uses integers only (since Pebble emulates floating point math)" [1]

Pebble has a soft-float implementation? So is this integer sqrt approximation is faster than the soft-float one?

[1] https://github.com/PebbleSeizureDetect/PebbleSeizureDetect/b...

I've used both, integer square root is typically much faster than floating point emulation.

Really almost anything is faster than FP emulation.

Awesome work, best write up of this type of project I have seen yet. But its been done before: https://github.com/OpenSeizureDetector/OpenSeizureDetector

Nice! This must not have been released prior to November of last year when I first authored my software. I searched around but couldn't find anything viable at the time, so I had to make my own.

This is amazing, and very important – this year I lost a friend to a seizure that might not have been fatal had they been found sooner.

Thanks for putting this out there - looks fantastic for hackers like me with partners who suffer from epilepsy.

If you are interested in wearable tech and epilepsy you should also check out the "Embrace Watch" [1] - It measures electrical conductivity across the skin of a person wearing the watch and is able to use that as a proxy for heightened electrical activity within the brain (which can be indicative of a tonic-clonic seizure). It pairs with phones to send SMS messages to nominated contacts as alerts.

It was developed out of MIT and is now being commercialized. I'm currently waiting to take delivery of this watch (from an IndieGoGo campaign last year), so can't speak to its effectiveness but I have high hopes, particularly since it doesn't just rely on motion. Motion detection is helpful, but by then it is too late for the person having a seizure as they have no warning to make themselves safe.

[1] https://www.empatica.com/embrace-watch-epilepsy-monitor

Yes, I am aware of this and my wife and I also backed the IndieGoGo. I hope it works as well as advertised! In the meantime, Pebble Seizure Detect is a decent stopgap measure. Empatica hadn't even been announced, when I first authored this software.

This is great.

There are commercial devices that work along similar lines, such as http://smart-monitor.com which is bought as a $20 per month subscription. But something open running on a platform like pebble is much more interesting.

This was supposed to be a comment expressing my fear of subscription-model hardware, but I looked at the site and - what the hell, take a look at [0] - aren't they just rebranding Pebbles? Most of the watches look very much like Pebble Time, and this one[1] is literally Pebble Classic (you can even see the standard Pebble status bar).

"Made by Smart Monitor, the SmartWatch is a patented, intelligent motion detecting and alerting wristwatch."

Yeah, right.

EDIT: The watch names suggest it too[2]. Also note that they're basically charging 50% more for the watch itself than the official Pebble shop, and that is before the activation fee and monthly subscription fee.

EDIT2: The more I search, the more I'm confused about them. They look legit, and apparently[3] they even had their own smartwatch prototype before moving to Pebble. But there's no mention of "Pebble" on their site, or in any materials related to them that I can find. Moreover, this interview[4] says they're planning to get this (presumably Pebble version) FDA-certified, which makes me wonder - aren't they going to find themselves in competition with Pebble itself? Ironically, both them and Pebble seem to be cooperating with Stanford[5][6].

Oh, and did I mention they claim to have a trademark on the word "smartwatch" itself?

[0] - http://smart-monitor.com/smartwatch/about-smartwatch/

[1] - http://smart-monitor.com/wp-content/gallery/gallery1/PW-watc...

[2] - http://smart-monitor.com/buy-now/

[3] - https://plus.google.com/+DanievanderMerwe/posts/BiNUbBgpi8B

[4] - https://www.youtube.com/watch?v=rcu8Ts9MFtY

[5] - https://www.youtube.com/watch?feature=player_detailpage&v=rc...

[6] - https://blog.getpebble.com/2015/12/15/health/

Interesting. I suppose its possible they made a deal with pebble. These things are never going to be sold en-masse, so why wouldn't pebble agree if someone came to them with a neat medical application.

Your search - site:smart-monitor.com pebble - did not match any documents

Pricks.

See also Embrace watch by Empatica which I think came out of a kickstarter or similar

https://www.empatica.com/index

Hrmm... Real time health monitoring could finally be the truly compelling reason to wear smart watches.

We need to wait 'till most manufacturers get bored with the current fitness frenzy and start building something meaningful. Or maybe independent app developers like the author of this one will beat them to it.

Maybe it's because fitness is a good 'safer' arena to enter and test sensors and algorithms before jumping straight into medical devices.. especially considering how the FDA feels about such devices.

This is really cool.

My wife died of a tonic-colonic seizure 11 years ago. Something like this might have saved her.

My previous partner used to suffer seizures of this type, this would have been incredible as often the only way I knew she'd had a seizure would be if she text me after she'd recovered, that was a terrifying thing considering they could hit at any time and any where.

How is development/testing done for this software? Is there a test data set of seizures to test against?

As sad as it may be, it seems from the README that the author was testing it on his wife, who suffers from those seizures.

Google returns several papers about detecting seizures with wearable accelerometers, but I haven't been able to find an openly available dataset. It would be great if there was one.

You are correct. I am aware of no public data set. I test it by watching YouTube videos (people often post videos of their seizures) and attempting to mimick the arm motions I see. Crude, for sure, but it's the best I can do.

I've heard several mentions of this over the years as one of the better use cases for devices. Sometimes it takes having a problem really close to your life to drive that devotion - I'm glad someone spent the time and effort to solving this.

Reply on news.ycombinator.com