The recent popularity of commercial wrist wearables has made it possible to study stress intervention systems in the wild, but there is a lack of pragmatic platforms for research prototyping and evaluation. We present an open-source toolkit for collecting raw photoplethysmography (PPG) data and modeling stress detection using Samsung Galaxy Watch, an off-the-shelf smartwatch. The feasibility of the toolkit for stress detection was validated against existing wearables such as Polar H10 and Empatica E4.
WatchPPG consists of
- a WearOS application on Samsung Galaxy Watch for collecting raw PPG data
- a real-time stress detection pipeline.
Below figure illustrates our PPG-based stress detection pipeline.
The pipeline takes a PPG signal and ouputs time and frequency domain heart rate variability (HRV) features commonly used in related researches:
Time-domain:
- the mean peak-to-peak (PP) intervals (mean PP)
- beats per minute (BPM)
- the standard deviation of the PP (SDNN)
- the root mean square of the successive differences of PP (RMSSD)
- the proportion calculated by dividing the number of interval differences of successive PP greater than 50 ms by the total number of PP (PNN50)
Frequecy-domain:
- low-frequency band (LF)
- high-frequency band (HF)
- ratio LF/HF (LF/HF)