Item — HN Reader

> A more real world case is that hospitals are starting to create out patient monitoring apps. Whenever a patient uses their blood pressure monitor, the hospital app gets notified of the new reading, sends that data to the hospital computers, a doctor might then review it, see there’s a problem and call the patient in.

I don't see how doctors could rely on this data. There are so many devices that may use HealthKit, and many of them are (for now) nowhere near as reliable as the equipment doctors would use for the same metrics.

Take blood pressure monitors, for example. If you shop around for a good one for iOS, even the best ones are still questionable, at best. Some people claim they're very accurate, while others say they're way off.

It seems we have a long way to go before doctors could rely on the data from HealthKit devices.

2 comments

I'd imagine that to begin with medical professionals will only allow this on approved devices (the originating device is stored with the sample). Given that certified devices have to work within defined tolerances, they'd then be able to know roughly how good the data they're dealing with. I guess there's always going to be cases of user error in equipment use, but that's part of what has to be considered when designing such a system.

The data may be wrong but couldn't the trends be useful? For example a heart rate monitor consistently off by -20bpm will still provide an accurate trend on heart rate over a period of time. I guess the problem here is that the data, even if wrong, would have to be consistently wrong by the same amount.

Your example is overly optimistic - if it was consistent like that, you'd just adjust it at the factory or calibrate it later on, and you'd have a perfect instrument. In reality, the data is all over the place, with a large spread. You might still be able to get some kind of weak trend out of it, if you knew the tendencies of that particular equipment. I doubt you could even take long-term averages and rely on them.

Take for instance an elliptical machine I own (it's about 4-5x the price of the cheapest ones, so I guess we can call it "mid-range"). The heart-rate monitor on it is almost worse than useless. Sometimes it will report 200 BPM when I'm not holding the sensors. Sometimes it will under-report for long periods (e.g. reporting < 120 in the middle of a session, where I'm fairly confident my actual heart rate is around 150). I would not recommend relying on these things much.

That said, I think the general idea is not a bad one; we could definitely use more data to analyse. It's just that equipment that gets sold to the public is pretty erratic at the time of writing.

I wonder why these instruments are so hard to get right. I'll be interested to see how well the sensors on the Apple Watch work considering it's essentially the official companion to the Health app. I'd assume that monitoring heart rate form the wrist should be quite accurate but it probably depends on how tight someone has the watch on and the exact position it is on their arm.

This. Trends are far more important for diagnosis than individual results most of the time. Trends in body temperature, blood pressure, blood sugar, etc usually tell a better story than one outlying result.