I'm building a night-time bruxism / sleep tracking wearable device


#1

Night-time bruxism is a medical condition where the patient unconsciously clenches their teeth together or grinds them sideways during their sleep. A lot of people have a mild form of this condition which can be harmless, but for some people, it takes on an extreme form and causes chronic headaches and dizziness.

The aim for this project is to measure the amount of night-time bruxism, and attempt to condition the user out of this behavior using an audio signal.

I built an early version of this prototype with an Arduino micro in 2013, with full details here: https://github.com/lucwastiaux/gc

I now want to build an internet-enabled version of this device, which will continously stream data into a time-series database (Influx DB) in an attempt to gather as much data as possible and determine the optimum biofeedback settings.

I have a prototype ready on the breadboard, which comprises the following components:

  1. Particle Photon
  2. Sparkfun LIPO power shield
  3. Sparkfun IMU shield
  4. Myoware muscle sensor, to be placed on the temporal muscle to detect clenching

I’m working with a PCB designer to integrate the above components into a small form factor. The aim of this post is to socialize and see whether people have feedback, about the project, or any suggestions.

Initial specs for the project: https://github.com/lucwastiaux/gc/blob/dev/gc2_specs/SPECS.md


#2

Awesome! Please keep us posted, I could see myself wanting to build one / buy one! :slight_smile:

Thanks!
David


#3

OK will do. For now, just wanted to report a stat on battery life. With the components listed above, i’m reading from the EMG sensor and the IMU every 250ms, accumulating data in a buffer and sending it over TCP every 5mn. The device is plugged in to a 850mah battery. During a 7hr night, the battery drains about 50%. This is very good as far as i’m concerned, given the battery has a tiny footprint.


#4

First of all…great project and hopefully it helps some people!
As for feedback, people love data and charts. I am a Hello Inc Sense backer/owner and have learned so much about sleep hygiene from their app. If the goal is to sell a device keep every tidbit of data. The term is cliche but Big Data is real and insights driven from it have intrinsic value. In fact it is widely believed that datasets are going to start finding their way on to balance sheets.

If there is any way to make the sensor wireless and non-intrusive for a user that sleeps on their side then you are really getting someplace!


#5

I am planning on exposing the data in Influx DB, using a Grafana dashboard. The primary focus will be the activity of the temporal muscle, which is indicative of night-time bruxism. The accelerometer and gyro data will help understand what phase of sleep the user is in. There will be a LOT of data to work with. I’m planning on developing a Web app around the data.

In silent data collection mode, the device shouldn’t interfere with sleep. However in biofeedback mode, the device will purposefully interrupt sleep in order to disrupt damaging clenching patterns (the sensitivity will be configurable). This means some sleep deprivation will happen, but for people with extreme clenching, this is preferable to the alternative (typically, permanent daytime headaches, extreme dizziness, pain while chewing).

My early device was supported by a neck strap and placed on the stomach, with electrodes going to the head. This means sleeping on one’s stomach was not possible. With this new version, sleeping on the stomach will be possible, but only on one side of the head, given the sensor will be on the other side.


#6

I’ve built Grindbit (http://www.grindbit.com), using the EMG sensor to get the data, then storage to the cloud. Idea behind is to compare that with other sets of data such as exercise, work location to see whether those have any effects on the grinding. This is third iteration I’ve went through, all the way from a smart mouthguard to this. I believe with data we can actually get some very interesting results.

Here are some of the initial image and prototypes


#7

I’m documenting the progress on the bruxism device on my blog:

https://sleeptrack.io/


#8

And here’s the first PCB using a P1 https://sleeptrack.io/2016/03/11/progress-update-pcb-prototype/
https://res.cloudinary.com/photozzap/image/upload/c_scale,w_1024/v1454817789/gc_website_blog/progress_update_1/sleeptrack_v2_pcb_front.jpg


#9

I wrote a post about the web interface which goes along with this device: https://sleeptrack.io/2016/04/14/first-look-sleeptrack-interface/
https://res.cloudinary.com/photozzap/image/upload//v1454817789/gc_website_blog/first_look_sleeptrack_interface/influxdb.jpg


#10

I gave up on using the P1 for now. I was running into difficult issues with the power supply, so I went a simpler route, with a headerless Photon. Device photos: https://sleeptrack.io/2016/07/09/phase1-hardware/
https://res.cloudinary.com/photozzap/image/upload/c_fill,w_1024/v1454817789/gc_website_blog/phase1_hardware/sleeptrack_phase1_components.jpg