Tracking My Tremor

I’m working on a Tremor Tracking project for the “Keating Memorial Self Research” activity in Open Humans. I’ve had this project going since last year, with many stops and starts, and I’m hoping to make some good progress in the next weeks and months. It would be great to have some companionship in exploring this specific topic. Essential tremor is one of the most common neurological conditions among adults. A common way to see your tremor is to balance a sheet of paper on your open palm, and see if it is clearly shaking. A more quantitative measurement can be obtained using the accelerometer on your phone and some analysis available in Open Humans. (You can see from the screen grab below that my tremor frequency is between 7-8 Hz.)

I don’t want to get into too much crazy detail here yet, since my goal is just to describe my topic and see if I can inspire others to join me. There is a personal tremor tracking log going back to June 2019 posted already on the QS Forum. I’ll be starting fresh here, but making use of what I’ve learned already and glad to support others who want to join.

1 Like

Since essential tremor is common, have you tried finding someone else with it to try doing this too?

I know using the physics sensor tools is a bit geek niche thing, but with a common condition it seems like surely there’s someone else out there interested in giving it a try. :smiley: It seems like it should be measurable in that data, somehow – there should be a way to measure “tremoryness” for folks!


I just come back from meeting the folks at the MakerLab at my institute. One of them is currently working on developing an open hardware device for tremor tracking based on the Movuino board. He’s currently beta-testing with patients with MS & Parkinson’s, but there’s nothing to be shared at this point I think.

Having said that: he recommended checking out PowerSense, which is an iPhone/Apple Watch app that allows to record the raw accelerometer data right from the Apple Watch. If you have an Apple Watch, this might be easier than doing the balancing of the phone?

Thank you @gedankenstuecke - I have an Apple Watch that I haven’t been using but I’ll take a look at PowerSense. The main challenge I’m having with my sensor-based measurement protocol right now is not with the action of making the measurement, which I’ve gotten pretty good at, but with doubts about how to turn the measurement into a meaningful tremor score. In other words: How do I calibrate the measurement? I can see that it is generally higher when I’m more tremory, and lower when I’m less tremory, so I’m sure it’s measuring my tremor. But beyond this very rough plausibility, I’m not sure it’s adding very much to my subjective measure, which I’m recording with the 1 Button instrument. I think a sensor based system would give me better resolution – but I haven’t gotten there yet.

That makes total sense. I think the combination of sensor-based measurement and the one-button qualitative recording might be really interesting for figuring out how to make use of the data? E.g. comparing sensor data recorded when subjectively feeling like it’s bad to random spot measurements?

Yes, I like that. I am gearing myself up for another 3 day “intensive measurement” phase.

@Agaricus You probably already know about Kate Rosenbluth . Might be interesting to see if your tracking could be combined with this someone.

1 Like

@sprague I wrangled at intro to Kate from a friend and just emailed her to describe my project and see about doing something together, it would be excellent to launch a PLR that used easy self-assessment/tracking of tremor. They are very clinically focused so this may seem too avant garde, but we’ll see. I’m also going to ask my doctor about the Trio. If it’s not too pricey I’ll try to get my hands on one.

This is an update on my project, cross posted from the QS Forum:

I learned a lot from the first phases of this project, and I’m now restarting based on what I learned. Some lessons:

  • My vulnerability to tremor varies during the day.
  • I can subjectively estimate my vulnerability to tremor (“tremory”)
  • I can see variation in the graph my sensor measurements.
  • I don’t have faith that my analysis of the sensor measurements (“tremor score”) is especially accurate as a measure of tremory, partially because I don’t understand the analysis well enough, and partially because I’ve seen variation in the score that is a lot greater than my experience of tremory.

My next steps are:

  • Use the 1 Button to track tremory. The challenge of measuring tremory (rather than tremor) is that the tremor is worse depending on the action I’m attempting to take with my fingers. Just tracking when it’s annoying doesn’t work. So I’m going to do a concentrated test using experience sampling to assess subjective tremory whenever I get a notification. I happen to be using imoodjournal (on @madprime’s recommendation) to do experience sampling of mood, so I’m going to make a tremory assessment at the same time.
  • Do a concentrated test of the sensor measurement at the same time as the subjective assessment, but only when this is convenient. I have to be at a desk.
  • Get help with the sensor analytics so I understand them better, maybe I can restore my confidence.

Right now I have a stupid problem, which is that I bought a new phone case that is more slippery than the old one. (I was cheap and didn’t want to pay Apple.) So I have to anchor the phone against a stop to keep it from slipping. Mechanically, I expect this to add stability and make it difficult (and for me impossible) to know the relationship between the new measures to old measures. I think that’s ok. I’m still in the idea generation stage, not in the prove ideas stage. In this next round of sensor measures, I’ll be starting from scratch.