Thread by @altini_marco

A question that occasionally comes up is whether you could use heart rate variability (HRV) continuously, during the day, to identify mental stress with e.g. a wearable

In my opinion, the short answer is no

Here, truly less is more

For the long answer, see below 🧵👇

Normally, when we talk about stress, we consider all sources (mental and physical), as physiologically, there is ~no difference (see slides 23-24 here: slideshare.net/marcoalt/heart…)

This means that you cannot separate mental stress from physical stress in HRV, in the "real world"

/2

This is also why HRV measurements should be taken in certain well-defined conditions, for example, first thing in the morning or during the night, so that baseline physiological stress can be captured in a repeatable context (more details here: medium.com/@altini_marco/…)

/3

During the day, many transient stressors will play a role, from diet to light exercise to mental stressors, etc.

Many of these transient stressors will have no long-term impact, while others will pile up and impact your baseline physiology

/4

Thus, morning or night measurements will be highlighting periods of high stress (both mental and physical) in response to strong acute stressors, or chronic ones, as shown here:

https://twitter.com/altini_marco/status/1442419577271504896

)

/5

While some devices try to use daytime data to look into stress, the nature of what is being measured (i.e. the autonomic nervous system activity) makes it so that in my opinion, the limitations are huge and practical applicability very limited

/6

For example, you could think of building a model in which your own HR / HRV during the day is compared only to your own history at the same time of the day (to account for individual variability and the effect of the circadian rhythm), and also only when you are at rest ....

/7

... however HR / HRV will react dynamically to everything: the time of your lunch, or the macronutrients in there, or if you had a hard workout, or were just chatting with a friend, your HR will also be high and HRV suppressed, without any movement or 'dangerous' stressor

/8

Plus, stress can be good. The goal should not be to get warned about any physiological change that resembles stress

Some stressors have a "negative" acute impact on the body but then lead to improved health (e.g. exercise)

Continuous measurement trivializes these aspects

/9

My point is that these metrics are useful when properly contextualized, e.g. morning or night measurements or even just before / after a stressor, during aerobic exercise, etc., but when looking more broadly at daytime data (automatically) with consumer wearables, not really

/10

Note that certainly larger stressors can be identified (e.g. sickness), but there would be too many false positives (periods in which you are not actually stressed), for this to be effective, in my opinion

/11

HRV is insightful in the context of mental stress of course, but continuous measurement and analysis without overinterpreting any minor change is a hardly solvable challenge unless the user is constantly in the loop (which would be alienating, talking about mental health..)

/12

Less measurements, more insights

One data point per day is really all you need to assess physiological stress and do something about it, whenever possible

Plus, this approach leads to a healthier relationship with technology, wrt continuous measurement (in my opinion)

/end