Hi!
My name is Nikolai, I am a Computer Science Student from Munich, Germany. I am currently working on my master’s thesis, for which I helped to build an Android stress tracking app. The App works with all Fitbit and Wear OS devices which have a heart rate sensor.
Get real-time feedback of your stress levels via an Android App
Check out how your stress levels changed during the day
Keep a stress diary and connect your computed stress levels with your real world activities
See your current stress level at a glance with our stress - complications (Wear OS only)
Measure your heartrates at adjustable intervals with a watch face (Wear OS only)
Fully customize the watch face with complications and colors.
To evaluate the work I’ve done, I am planning a user study for which I am desperately searching participants.
The study will be running from 17th July to 31st of July. Participants are asked to use the app and regularly track their activities and feelings. This helps us to improve the stress detection algorithms.
Thanks for your interest in the app!
For the instant feedback we are using simple HRV calculations. We are trying to differ between positve stress (active skiing) and negative stress (fight with girlfriend) by training a machine learning classifier on your heart rate data: For that we need labeled data in form of a stress diary of as many users as possible.
You can access the predictions of the classifier via a website after you start using the app. We plan to release this feature in the Android App as well but for now you’ll have to switch to your browser to see whether you were experiencing good or bad stress.
Hi. What is the difference in HRV between good and bad stress and how long do you need to get a reading. Are you using a spectral analysis of the ECG? Marc
As the HR measurements of the Smart Watches are not as precise as of a clinical ECG we’re currently only using analysis of the time domain.
We are training the classifier with multiple attributes of the HR Measurments including the mean RR interval, the HRV Score and many others.
By labeling the data you can teach the classifier how your heart rate looks when you are positively or negatively stressed.
When there is enough data, a pattern for eustress, distress and rest is created, which can later be used to compare against your measurements. You can also explore a graph of this progress in our web app.
We calculate HRV on 10 minute windows, so that’s how long it takes to get feedback whether you’re stressed or not. The classification between good and bad stress takes a little longer (depending on the server load). Users manually upload their data to our server by pressing a button, so there is no live feedback on the kinds of stress you experience.
I would be interested in testing. How do you plan to collect diary data. Is there an android app that does that, maybe with a simple number of frown faces?
Hi Ivan,
thanks for your interest.
Please check out this website for more information about this study and how you can apply: http://stila.pms.ifi.lmu.de/news.html
You can track your activities in our Android App or in our Wear OS App right on your watch. The App is currently in Beta but you will gain access when you participate in the study.
It’s up to you how detailed you want to track them, we offer various attributes like mood, posture, drug intake etc. which can all reflect on your heart rate and might allow us to improve our algorithm.
The App then shows you a daily overview of your tracked activities with your stress perception on the watch and the phone.
You can also see this in the screenshots I linked in the first post.
It looks like the studies been over for some time now, but the webpage is not updated with the results. When do you plan on making the results avaliable?
