Have you ever experienced an episode of acute stress in the context of your work? If so, how long did it take to feel a full recovery?
Yes, on a daily or weekly basis I feel the need for a break, and for that I have a few escape valves, such as my love for classic cars. I spent a long time looking for a hobby like that so as not to take the stress home with me.
Would you feel comfortable using a wearable such as a band while carrying out your work?
I don’t use one regularly because I never really got used to it, but maybe that was just a matter of habit. The charging and other aspects introduce a series of demands that, in themselves, are one more element of stress.
How do you see the prospect of getting a weekly report of your psychological activity in terms of focus and stress? Would it be a good complement, or just a useless feature?
I think it would be important to have that report. In the things I usually do, I only deal with productivity at different intervals rather than weekly. As for my usage profile, I’d prefer to keep them turned off on my phone.
To what extent can we trust our sensors and the reliability of the data collected?
Working in the field, I’m always very skeptical about how a sensor performs, and we’re very critical depending on how things are advertised. Even with more reputable manufacturers, what we get may not be very useful. With studies backed by a supporting population and correlation between sensors, however, you can trust it.
The reliability of an optical sensor, in turn, is very much related to the stability of the pulse, which is why it’s important to use an accelerometer.
As for the EDA sensor, there are many variables that change and it needs quite a lot of adjustment, but since its activity is linked to the sympathetic nervous system, together with heart rate variability, this sensor helps provide concrete and reliable information about a person’s mental state.
There can also be crosstalk between sensors, and you need to be careful with that.
What are the most common ways to process sensor data?
In that kind of approach, among the various techniques used (high-low filters, algorithms, etc.), the most common is machine learning to classify the signals. We usually have a different point of view from clinicians; we don’t follow a standard. We use several metrics, and what we do is put everything into the algorithm and train it based on placing someone in a more fun or more boring activity.
At the start it’s harder, and we rely more on the person’s own report after that task. We also try to cross-reference with environmental and context data (how the person felt, the conditions, etc.). We use metrics geared more towards self-knowledge: based on the signals, we show the person what they were doing at that time of day. This has individual implications for measuring their wellbeing (we didn’t have many concrete cases that led to a substantial change, drawing that from the person’s statistics).