Hacker News Reader

Then it's in line with the accuracy of current fitness watches on the market that can't see my 200m sprints, think I'm taking steps when I'm eating with a fork, and undersample + oversmooth my pulse.

Gathering sketchy blood pressure data all day in the hopes of smoothing it into intelligence just sounds like the state of the art for wearable fitness gadgets.

Yeah, except... Who wants to be the guinea pig here? There is a difference between "oh, I did 7000 steps instead of 6000" and "oh my blood pressure is 200/90 instead of 120/70".

We are getting so used to poor software just for the sake of "innovation" (release fast) that I am wondering what's the difference between buying a fake watch and a real approved one - like this one from omron.

There are four different scales here.

One is the current generation of sleep trackers or Bluetooth toothbrushes (I own both, just for curiosity’s sake and amusement) which at best tell you only what you already know, and often not even that.

Next is the coarse 6K/7k step case you talk about. The measurements on the gym’s treadmills are similarly arbitrary. That kind of gamification is useful to get people moving and as you say accuracy isn’t important.

Next is the gross trend...if you’re in a pattern and deviate then it could be a sign of something and could be that you put the device on loosely or that it is dirty.

Which is the last kind: even though I have the watch i still use a pressure cuff, though not often, and even occasionally exercise with a heart rate monitor and even a pressure cuff. They suffer from the worst sort of measurement problem: inconvenience. Just as “the best camera you have is the one with you” the same applies to health monitoring.

As for the importance of health monitoring...how often do you look at your poo? Even in Germany where many toilets are equipped with a “shelf” for inspection, an informal (and amusing) survey I conducted revealed that most people never looked even when they weren’t feeling well.

> Next is the coarse 6K/7k step case you talk about. The measurements on the gym’s treadmills are similarly arbitrary. That kind of gamification is useful to get people moving and as you say accuracy isn’t important.

I wish the industry had converged on a more abstract metric name than "steps". Just as I shudder at how many fitness trackers now have been moving to metrics called "calories" or "active calories", which seems like an over-correction away from "steps" to a more dangerous alternative. They imply a physical specificity that isn't that clear cut a direct correlation.

> Just as “the best camera you have is the one with you” the same applies to health monitoring.

This does seem to generally be the case. Where most people never bother to check their blood pressure at all, even a false positive is more awareness of their health and its implications/consequences than they likely had before. So long as warnings are (as they usually must be, given laws and liabilities) highly couched in the usual disclaimers to seek further medical advice.

It's amazing what data we are slowly accumulating from our wrist-based tricorders, even if this is all still very early days in figuring out how useful any of it is, how accurate we can get any of it.

At least "steps" are understandable, and the data are roughly correlated.

It could be a lot worse: I own several sleeping bags for winter backpacking. Each has been variously rated by its manufacturer as 10 C, -10 C, -30 C etc. There is no standard for this and in fact there is no correlation with performance: the best one for really cold conditions is rated -10 C.

I think Fitbit and Apple at least tried to have the number of steps taken by an "average" user registered as about the same number all the time.

> At least "steps" are understandable, and the data are roughly correlated.

The point though is that it is very rough, and they may as well have always been called "fitpoints" or "beans" or "gimmicks" for what good it does calling them "steps", especially because of that confusion both inside the industry and out that A) devices have moved from being directly pedometers at the foot to being indirectly pedometers at the waist or now the wrist, and more critically B) that they were always meant to be more of an overall aggregate activity score and the one specific physical action of a "step" never the end goal in measuring the metric anyway (it was just the easiest metric to measure in the early days of pedometers, and being called "steps" as much a historic throwback than an accurate name even in the early days of pedometers).

Whether or not the metric is consistent today, it's at least somewhat in one direction or another inconsistent to call it a "step" and an abstract name would have been more fun and also likely avoid more confusion than the "rough correlation" suggested today.

The shelf is, in fact, to save you from water splashing your cheeks, not for inspecting...

I like when mine tries to guess if I'm cycling. It's usually about 20 minutes into the ride that it figures it out and starts the counter.

If you're in good health, your watch may actually be doing a good job measuring your heart rate. When you go cycling, it just takes that long for your heart rate to increase significantly enough to trigger the watch's activity detection software.

The way I see it, the problem isn't a question of accuracy so much as the watch trying to guess activities it has no business guessing. The watch should focus on things that it's good at: gathering a reasonably accurate consumer grade heart rate and maybe showing notifications or the time if it has a display. Leave activity detection to something with more situational understanding.

I don't know what you're using but modern GPS watches can absolutely track 200m sprints with decent accuracy. Adding a foot pod sensor can also help improve accuracy.

>Now, to put this watch in perspective - this is a wrist BP monitor that measures at will and claims clinical accuracy, which is hard to get from a regular wrist monitor under ideal conditions.

I don't know if this is how they do it but if you are wearing the watch for longer periods, and the measurement errors are random, it is plausible that you can confidently calculate an accurate overall BP out of many readings done in the background.

From a random company I'd be sceptical but given that it is Omron.. well, I am still skeptical but a bit less so.

> out of many readings done in the background.

You have to hold your wrist at the height of your heart for it to take readings, so it's not happening in the background.

The errors are most certainly not random, especially in hypertensive patients who are most likely to purchase these devices. Hypertensive patients can show large variability in successive BP measurements just due to arm position.

I assume "clinically accurate" is a completely meaningless term, like "clinically tested" or "clinically proven" on half the items at the drug store.

I'm having very hard time believing that this device is in fact "clinically accurate."

Don't they have to be certified by the FDA to be marketed for that purpose? I recall someone at Apple talking about this sort of thing with the Apple Watch monitoring different health functions.

I think it varies.

This should be an interesting read (not entirely related but somewhat similar): https://www.fda.gov/inspections-compliance-enforcement-and-c...

The Apple Watch required FDA clearance (though this was only the 4 with the ECG feature). Since the EU has required approval even for some phone apps for fertility cycles I think a fitness watch that says it provides clinically accurate blood pressure readings may fall onto some radars.

I don't think the FDA, and especially EU regulators, look upon consumer devices that replicate medical functions lightly, since these can burden healthcare professionals and result in misdiagnosis/self-diagnosis.

See my comment elsewhere: I could not find filings for this device

Just because they haven't figured out how to back out the pseudorandom variations to your blood pressure doesn't mean it will never happen. And getting these devices into people's hands is the first step in that direction. At some point, we'll be able to take data from continuous blood pressure readings (regardless of arm position) and reference them against time of day, heart rate, activity level, accelerometer readings, diet, etc. And on the other side, we'll have an idea of, not just your blood pressure at rest (which has the virtue of being consistent but isn't the best predictor of heart attack risk), but your blood pressure during daily activity and intense exercise.

Complaining because we're only part of the way there is counterproductive.

Yep. Blood pressure monitors' accuracy goes like this, top-to-bottom, most to least accurate:

  Mercury sphygmomanometer
  Aneroid sphygmomanometer
  Digital meter (arm-cuff)
  Digital meter (wrist-cuff)
  Digital meter (finger cuff)

The mercury sphygmomanometer is the kind with the upright scale on a mercury column that sits on a table-top and is operated with a hand-pump and a stethoscope. The Aneroid sphygmomanometer is also hand-pumped and used with a stethoscope, but it's got an analog dial and it does not have a table-top component.

The watch we're talking about is a digital meter with a wrist cuff, so it should be less accurate than a digital meter with an arm-cuff but more than one with a finger-cuff.

Since I own an Omron wrist BP monitor and it, as a dedicated medical device, is nowhere near the precision of mercury sphygmomanometers that doctors use, I seriously doubt their claims of "clinical accuracy" for a watch... it's probably just something that marketing department thought to sound cool

Yea, it seems like it would be some special breakthrough as the systolic (distyolic? i get left ventricle open closed confused too) number in mmHg is how much pressure it takes to cut the circulation off at your arm.

Good enough accurate? Maybe. As good as cuff, pretty much flat out no.

My thoughts exactly. I'm not impressed with the accuracy of Omron's wrist BP meters, so why should I trust this thing? Also I'd expect the battery to die after about two readings; measuring BP requires a lot of energy.

> elevated diastolic (lower) pressure

Somewhat tangential, but what is this usually an indicator of?

My pressure will sometimes be 115/85, and the 85 seems kind of high. Just curious what causes this and how to fix it.

There are 2-3 known causes (don't remember them offhand, sorry), but they account for like 10% of all cases. Causes of the remaining 90% is unknown.

85 is not high, but once it routinely crosses 90 they usually prescribe a low-dose medication to knock it down. This is to reduce the risk of excessive blood vessel wear.

Read the article. The device mentioned here has to be held over the heart to measure blood pressure. It doesn't actually work while on the wrist, apparently.

As such, this is pretty lame.

The device mentioned here has to be held over the heart to measure blood pressure.

Elevating the device to to the level of your heart seems like standard practice. Every home blood pressure monitor I've ever used requires this. And it's also the way my doctor does it in his office with both electronic and the squeezie bulb blood pressure thing.

>squeezie bulb blood pressure thing

A sphygmomanometer?

https://en.wikipedia.org/wiki/Sphygmomanometer

It's the standard requirement for all wrist devices, with arm-band you can hold your hand in any relaxed position

Yep - most nurses miss this

I would be shocked if a nurse made such an error.

It’s also important that you sit quietly for a minute or so and with both feet on the ground before taking the reading

it works on the wrist, it only takes the reading when you put your wrist (and the watch) over your heart. It actually forces you to literally put your wrist at heart level.

Our Omron arm cuff one reads 20 pts too high. We didn’t realize until a doctor had us bring it in.

I have an exceptionally inaccurate Omron monitor as well. If they can't get it right with a full up cuff I really have doubts about a watch.

Mines quite accurate against the ones in my regular clinic - and with BP its more the change you want to track.