5 min read

Wiggly health, wiggly graph

I've been in a crash recently, so I figured it was probably worth putting a bit of money into a subscription to Visible. They're a startup who are trying to do the fitness band experience, but for people with energy limiting illnesses for whom optimising their physical fitness consists of managing walks to the shops not runs along a marathon. When I feel good I can get a little careless with thinking through my level of activity. And then I crash and regret it - so hopefully this subscription will help me manage my level of activity a bit more closely.

I'm not gonna review the app or the experience using it just yet - I've not used it for long enough - but using it is bringing me back to the early days of trying to get similar kinds of data from fitness trackers. I bought a Polar chest strap & wrote a little app to collect the data & graph it. Specifically, I was interested in this measure "HRV". This stands for Heart Rate Variability, and it's a ratio between the low frequency and the high frequency variations in timing between the beats of the heart. These variations become more regular at times of high stress, and looser at times of low stress (when the parasympathetic nervous system is activated). And when I say stress here I don't just mean work stress, but more: high stress is activation of the sympathetic nervous system, fight or flight mode. And low stress is activation of the parasympathetic nervous systems, rest and digest mode. (Can both modes be activated at the same time? Yes, the body is very complicated).

And Visible tracks this - I'm not sure if it uses the heartrate tracker to do so? Hopefully! It can read these things out, although the UI doesn't expose it on a continuous basis. Anyway, Visible tracks this, at least using the iPhone camera as an ersatz heart rate monitor, and it can be used as a measure of how well rested you are, compared in similar situation and similar times of day.

Anyway, that's not what I wanted to talk about today. What I've actually been thinking about is what the POTS[1] specialist I managed to see a few months ago said. Which is that POTS should not be seen as a condition characterised by a increase in heart rate upon standing, while blood pressure remains constant, as the clinical criteria say. But a condition where standing causes blood pressure to vary erratically on very short timescales, and the rise in heart rate is merely a byproduct of that. Now in both cases, it's the body struggling to maintain equilibrium in the face of having to pump blood all the way down to the floor and back, but it's characterised differently.

Now, the question is: why do we use the first definition and not the second one [2]. Well, he said that the big reason is that it's easy to measure heart rate (eg with a stethoscope), and easy to measure blood pressure on the scale of a few minutes (eg with an arm cuff), but hard to measure blood pressure changes on the scale of milliseconds. To do that, you need the special £25k machine that they have in the hospital, which clips on your finger and makes a nice wiggly graph of blood pressure changes. And that's fair enough, I can understand that dynamic.

But I've been on this wearable sensor chain of thought just now. And so I wonder - how does that expensive machine work? It's not invasive, it's just a thing that clips onto your finger and I think shines a light onto your finger. Much like a pulse oximeter, and those can be had for like a tenner. Is there some fancy tech involved? Could you find a way to make a... well, maybe not medically approved, but a consumer grade reliable blood pressure variability monitor?

Anyway, a little bit of digging suggests (but does not confirm) the technique "photoplethysmography", which is what's used for getting blood oxygenation readings.[3] It detects changes in blood vessel size when the heart beats, and therefore can be used to get an indication of blood pressure. And indeed there's a growing category of "cuffless blood pressure" monitors, in exactly this kind of consumer market.

So now I have some questions:

  • how good are these cuffless BP devices? is it worth getting one? can I read the data from it?
  • is the sensor data good enough in order to find this kind of sub-second variability in blood pressure that this specialist described to me? this suggests it might be, theoretically
  • can the sensor data be corrected such that these readings will be any good for a wearable sensor, in times when the person using it will not be at rest? my understanding (eg from the above link) is that this is hard
  • I mean for that matter, how hard is it to pull out the BP signal from the raw PPG data? when I go looking for this I see all kinds of stuff about random forest feature extraction and gray wolf optimisation[3], which suggests that it is hard. But also a subject where I maybe could contribute some muscle if I wanted to.
  • can I find a paper which goes into more detail about the relationship between POTS and this kind of BP variability? what's the right jargon for the wiggliness he showed me? how mainstream is this view? what impact does this have on hypothesized underlying mechanisms for POTS?
  • for that matter, this wiggliness in heart rate over short periods of time seems pretty reminiscent of that variation in heart beat timing in HRV. I saw a little bit of stuff in looking this up that suggested that it might also be indicative of SNS/PNS activation, maybe in a way that's clearer than HRV? Maybe this would have broader use for people who are not POTS-y?
  • would this direction of work be materially useful in providing clearer diagnoses for POTS patients who do not currently get access to the £25k machine? would this work help perhaps give an indication of varying degrees of POTS-y-ness over time, and therefore help guide sufferers in knowing how bad they have it of a morning? generally I see value in letting people see this kind of data for their own experiences, and then maybe generating new hypotheses based on that.
  • would this kind of work help strengthen the community-minded spirit that I so often see in sufferers of these kinds of badly diagnosed energy limiting illnesses? there's a real solidarity within these communities, galvanised by the surge of sufferers of long COVID. is this something that would one day be in Visible? or is this an example of the kind of deeper, nerdier, less easily explained stuff that could be in a version of Visible that wasn't made by a revenue-optimising startup? how do I feel about a startup aiming specifically at this community?

Lots of questions! I don't have good answers to them, and I think I have enough videogames to make that I don't think I'm likely to devote much time immediately to answering them. But a question shared is a question halved, so here's a chain of thought that someone else might want to carry forward.

[1: Postural Orthostatic Tachycardia. Which is basically what I have. I summarise it as "I get stupid if I'm upright too much".]

[2: It should be noted that you don't actually have to press up against the skin to do this - you can actually detect heart rate from distant video footage by analysing it to reveal minute variations in skin tone as the heart beats. Yep, you could do this with, like, a news interview with a politician. Weird and creepy tech!]

[3: My cog sci background means I have heard of random forests, but "gray wolf optimisation" is new to me and seems like an unnecessarily badass name for an algorithm.]