Measuring Stress with Wearable Technology in VR
August 16th, 2024
As immersive as Virtual Reality (VR) experiences are getting, they still lack one vital component- adaptability. Research has shown us that different people can experience different levels of immersion and presence while engaging with the same VR content. So, while I might scream my lungs out when a zombie decides to jump out of nowhere and run towards me, you might remain completely unbothered. And who knows, you might be a zombie survivor who, through a lot of trial and tribulation, has gained enough experience to know that screaming is not necessarily the best survival technique. But wouldn’t you still want to feel the same amount of endorphins as I do when I am so immersed in the zombie VR scenario? Let’s assume -I know that’s not very scientific of me- that your answer is YES. Then, it’s up to us to figure out how we can adapt the VR experience in a way that is equally immersive for you and me. One way to do this is to track your real-time experience with physiological sensors and use the data from those to alter the virtual environment for you. In this scenario, accurately measuring your physiological signal in VR becomes vital. Vital, and yet, not something commercially available. And this is where our quest started.
The first question we had to answer was which physiological signal we should target. Since one of the quickest stress responses can be read in the measurement of electrodermal activity - EDA (i.e., how your skin's electrical conductance changes, often due to sweating, which can indicate emotional arousal or stress) we set out to integrate EDA measurement into VR experiences using the EmotiBit sensor box (see the image below).
Image retrieved from https://www.emotibit.com/
The next challenge was deciding where should the EDA measurements be taken. To answer this question Rune Vandromme conducted a literature review to understand the optimal sites on the body for EDA measurement during VR experiences. Her research confirmed that the palms were an optimal location due to their high density of eccrine sweat glands (approx. 600-700 glands/cm²). Rune also consulted with stakeholders, including VR game developers, to ensure that the sensor placement location would meet all practical needs.
Next, we designed a couple of add-ons for the EmotiBit that would allow it to work seamlessly with VR controllers (in our case these were the HTC Vive Focus 3 controllers). The design process followed the principles outlined in Research Methods for Product Design (Milton and Rodgers, 2023). We began by immersing ourselves in the VR environment, using the "try it yourself" method to identify potential issues. This hands-on approach helped us understand the context in which the measurements would be taken and highlighted the challenges posed by the VR controllers.
In the next phase, we generated several ideas for integrating the EmotiBit with the VR controllers. We created prototypes and conducted user testing to evaluate comfort, usability, and signal accuracy (i.e., how much data loss we encounter). The tests revealed that hand size significantly affected comfort, leading the team to develop two different prototypes: one for smaller hands and one for larger hands. To be more precise, we measured the hand size of our participants’ hands and compared it with the hand size of a sample of Belgians aged 18-65 (Volwassenen lichaamsafmetingen, 2024).
Scetches of one of our initial ideas.
User testing involved placing the sensors on different parts of the hand and monitoring the quality of the EDA signals during VR gameplay. We discovered that good and consistent skin contact was essential for accurate measurements. Based on these outcomes we adjusted our designs, eventually selecting a flexible, compressible material called VarioShore TPU Prosthetic for the final prototypes. This material improved comfort and could withstand the rigorous cleaning required in public VR spaces or research facilities.
The two prototypes with different placements: A - places the sensor box on the hypotenar site; B places the sensor box on the palm site.
In the end, through testing and iteration, we developed an add-on for the EmotiBit that enables accurate EDA measurement during VR experiences. This innovation opens up new possibilities for using VR to monitor stress in virtual environments potentially opening the door to creating adaptive user experiences that can make each visit to Zombie land equally immersive for me (aka the avid screamer) and you (the veteran zombie hunter).
Finally, as much as I like using VR for its entertainment factor, I want to stress that this add-on can be used in many other contexts such as real-time stress detection in mental health apps, stress management serious games, exposure therapies and many more.
This research was conducted by Rune Vandromme, Jelle Saldien, Jamil Joundi and Aleksandra Zheleva.
For the engineers and all other curious minds - you can read the full research report here.