Seizure Alert Alarm Clock

Hello All,

I am epileptic and wear the Empatica Embrace watch, which detects seizures based on electrodermal activity and accelerometer data, then notifies the Alert app on the patient’s phone via bluetooth, which sends phone calls and GPS data over SMS to caregivers. Although very little is known about its underlying causes of Sudden Unexplained Death in Epilepsy (SUDEP), it has been reported that keeping a person having a seizure responsive immediately after their seizure can help prevent the phenomena, which kills more people per year than SIDS. Very recently, a case study was published in Neurology that detailed a case of SUDEP in a teenager wearing the Embrace watch. His father received the SMS and phone alerts but did not respond for fifteen minutes. This case study indicates how critical a quick response to this seizure alert can be, which inspired me to create an Arduino-powered seizure alarm that is more powerful than the standard phone alarm. The advantages of this alarm would be that it negates any issues of the caregiver’s phone being on silent, or the caregiver not being awoken by the alert. This will be my first Arduino project, but I am thoroughly committed to making this project a reality. I’ve done a decent amount of research on the hardware side, but have barely touched the software aspect of this project.

The general framework of this project would be that the watch detects a seizure and sends an alert phone call and SMS to a registered caregiver. Either the phone call or the SMS triggers a bluetooth signal from the caregiver’s phone to the Arduino at their bed-side, which activates a speaker and a mattress-shaking motor.

On the hardware side, I figure that the Arduino Uno should work just fine as the microcontroller for this project. In terms of input, I’m not sure whether the Bluefruit LE UART Friend or the Bluefruit LE SPI Friend
would be more appropriate, guidance here would be much appreciated. For output, I’ve cannibalized my Sonic Bomb alarm clock, which has an 8 ohm 1 watt speaker, and a purpose-built, mattress-shaking, 12v motor that attaches to the clock via a barrel jack. For the motor, I plan on using a TIP120 transistor and an external power source, but if anyone knows of a safe way to safely power the 5v Arduino and the 12v motor though the same power source, that would be ideal. Additionally, the female end of the motor’s barrel jack is mounted to the clock’s PCB via through holes, so I would like to de-solder it and use it for the project to make the motor replaceable if this would be possible. For the speaker, I’m unsure if I’d be able to use a Adafruit Wave Shield and still mount the bluetooth board and the motor components. Would a Wave Shield give me any advantage over a 2.5w Class D Amplifier and a MicroSD Breakout Board? Do I even need a MicroSD card reader if I’m just playing single frequency alarms rather than songs? I feel like these are the major components I’d need. Please let me know if I’m missing anything huge; I still need to learn more about selecting the right resistors.

On the software side, I’m hoping that I’ll be able to send the bluetooth signal from the caregiver’s phone to the Arduino using IFTTT and the Bluefruit app, but I’m not sure if it’s possible. If it isn’t, I’m not sure what I’d have to do short of writing my own app, in which case I’d be kind of screwed. If I can get this part of the project to work, I figure that getting the output to the motor and speaker working will be a layup. Still uncharted territory for me, but I’m sure I can learn this part quickly.

Thank you to anyone who can help with any aspect of this project! I feel like I might have bitten off more than I can chew here, and I might be asking some fairly basic questions, but if you can even point me in the right direction of what to read, I’d be very appreciative!

griffinvanhorne: The general framework of this project would be that the watch detects a seizure and sends an alert phone call and SMS to a registered caregiver. Either the phone call or the SMS triggers a bluetooth signal from the caregiver's phone to the Arduino at their bed-side, which activates a speaker and a mattress-shaking motor.

Keep in mind that you're still relying on the phone. It has to be turned on and not have a dead battery. It has to be able to actually receive the call and/or SMS message, and be able to respond appropriately (with the Bluetooth signal) when it is received. I have seen it happen where a person with a smartphone had to borrow a phone to be able to call out. The reason for this is that his phone was doing a system update, and there was nothing he could do to cancel or postpone this update.

Also keep in mind the possibility for other reliability issues. How much do you trust the software and/or hardware you are using? If you come up with something 99.5% reliable, what happens the other 0.5% of the time, and who will be held responsible for the consequences when it happens?

To be clear, this project is in no way meant to replace the existing alert system, merely augment it. I would absolutely love to have a project that works 100% of the time, but any amount of success provides an advantage over the current system. However, you're totally right; there are definitely still issues arising from the phone connectivity. An alternative could be using the FONA 800 Shield to have the Arduino directly receive phone calls and SMS. Would it be possible to have a phone call or text from a specific phone number trigger an Arduino command via the FONA?

I would think that backup/additional systems are a good idea. And if your idea gets help to you even a few seconds faster, all the better. Curious though, does the watch have any Wi-Fi capabilities? Or is it purely connected via cellular?

Thanks for the support! The watch itself doesn't have wi-fi connectivity, and to the best of my knowledge, the alerts can only be sent if the patient's phone has cellular connectivity.

I recall Arduino is explicitly not to be used for medical applications. Check on that (legal liability issue).

Secondly, the existing watch has the detection part well under control, as I understand from your post. Don't mess with that, I bet it's not trivial and easy to break, instead look at the alarm side: how to improve on the alarm, as that's what you're concerned about. That may be changes to the existing Android/iPhone app, or adding some noisy hardware or so.

As for hardware, I suggest you get one of these, if you can: