0

A lot of people wear smart watches and other fitness trackers these days. All of which measure at least the wearers pulse, usually a lot more data – even automatically recognizing type of exercise. Some claim to analyze stress levels directly.

The measured data is usually stored on servers of the manufacturer and does provide a large amount of data about individuals across regions, even whole countries.

I know that analysis of data like this is common practice in other fields. However, I’m unsure if the available data from the fitness trackers could provide clear enough pattern for any usable results.

Data access issues aside, to examine the concept of the idea, can the pulse histogram data (for example, of a large amount of people) be used to identify stress and anxiety changes – possibly correlated with events in the given region to analyze their impact?

12
  • Welcome to this community! I recommend doing another round of proofreading since there's a few typos here. Also, it's an individual that wheres a watch and has a biological pulse, not a country, so perhaps you can ask if a smartwatch can measure an individual's anxiety, then extrapolate yourself about trends in a larger population of people with smartwatch data. Mar 1 at 14:41
  • 1
    Please edit the question to limit it to a specific problem with enough detail to identify an adequate answer.
    – Community Bot
    Mar 1 at 14:42
  • 2
    I'm sorry, that the focus of the question was not clear enough, will edit it.
    – bennn
    Mar 1 at 21:09
  • 1
    @bennn it's actually a fascinating research question and a great first question by a new user. As comments point out it needed some polish, and honestly sometimes people reflexively downvote question that are simply out of the ordinary, so don't worry about those. There are almost 200 Stack Exchange sites to choose from and one that you might find helpful is Open Data SE, for example Location-based sleep dataset though it's currently unanswered. I'll give it a bump
    – uhoh
    Mar 1 at 21:47
  • 1
    This question lacks prior research. Do those devices actually measure stress and do so accurately? Without knowing the answer to that question the only possible answer to this one is "maybe." So you need to find evidence (or lack thereof) that those devices can actually measure stress and add it to your question.
    – Carey Gregory
    Mar 1 at 22:18

1 Answer 1

3

Tiwari et al, 2019 found poor sensitivity and specificity for detecting stress using multiple features from a fitness watch: sensitivity 0.53, specificity 0.56, and similarly poor for anxiety: sensitivity 0.44, specificity 0.63.

Of course, this measures stress/anxiety in individuals, which is not quite the same as measuring it in populations.

I did not find any study that attempted to do what you describe: estimate population levels of stress or anxiety and correlate this to events, but probably the first step would be to do a validation against some gold-standard assessment.

In research and publishing it's not so much about what is possible, which seems to be a binary "yes it is possible" versus "no it is not possible", and more about impact of a given result. If you show a correlation, will people believe it? They're probably more likely to believe results that confirm their prior expectations, and disbelieve results that oppose them. For a measurement modality without a track record of success, that's the correct behavior. With more use and more evidence that a given measurement accurately reflects reality, people will begin to trust those results more than their preconceived ideas. I don't think this is something you can know or predict without having actually done the research.


Tiwari, A., Cassani, R., Narayanan, S., & Falk, T. H. (2019, July). A comparative study of stress and anxiety estimation in ecological settings using a smart-shirt and a smart-bracelet. In 2019 41st Annual International Conference of the IEEE Engineering in Medicine and Biology Society (EMBC) (pp. 2213-2216). IEEE.

Not the answer you're looking for? Browse other questions tagged or ask your own question.