Well clearly this was a waste of time... http://9to5mac.com/2015/03/20/apple-watch-development-health-fitness/
Got a tip for us?
Let us know
Become a MacRumors Supporter for $50/year with no ads, ability to filter front page stories, and private forums.
Heart rate updates after 1.01
- Thread starter Pagemakers
- Start date
- Sort by reaction score
Thanks for the clarification.. I get what you are saying. I hope you are right about Apple integrating sleep tracking into the AW.
But for now, I am going back to the Basis Peak (just won an auction on EBay to replace the one I sold) ... I just miss the wealth of data it provided about my wellness, training and recovery 24 hours a day. I just can't get that from the AW yet. I will hold onto it for now in hopes either Apple or 3rd party developers (if allowed) unlock its potential.
Ha! Maybe Apple shouldn't have hired the Nike FuelBand guy?
I'm just flat out confused. I consider myself to be relatively literate with gadgets too.
Hey guys, I found this interesting article: http://www.engadget.com/2015/05/30/apple-watch-heart-rate-update-explained/
So I guess Apple meant to make the readings irregular.
So I guess Apple meant to make the readings irregular.
I can see the logic behind the whole idea. the infrared sensor used passively isn't as accurate as the green led used during workouts, so they're trying to reduce inaccurate readings by not allowing them to be taken during motion.
HOWEVER,
Let me just point out that previously, the materials stated that if the infrared led doesn't successfully get a reading when passively measuring heart rate, the green led would kick in to finish the task.
So considering the green is far more accurate than the infrared (thus why green is the only one used during workouts) this whole "movement produces inaccurate readings" thing is invalid.
Rather than nerf the passive infrared sensor, why not just tweak the conditions under which the green takes over to get an accurate reading?
HOWEVER,
Let me just point out that previously, the materials stated that if the infrared led doesn't successfully get a reading when passively measuring heart rate, the green led would kick in to finish the task.
So considering the green is far more accurate than the infrared (thus why green is the only one used during workouts) this whole "movement produces inaccurate readings" thing is invalid.
Rather than nerf the passive infrared sensor, why not just tweak the conditions under which the green takes over to get an accurate reading?
Heart rate updates after 1.01
I've followed a lot of this thread but not all of it, so hopefully this doesn't sound stupid. I can't fully understand their reasoning here, but maybe after the watch has enough data from accelerometer AND heart rate it decides it can get away with accelerometer only (saving battery)? With occasional extra readings to keep up with increases/decreases in fitness?
Maybe Apple is satisfied that they know your moving heart rate based on accelerometer and have no need to measure it. But they are interested in the resting heart rate as that is a great measure of fitness.
Would unpairing/repairing the Watch delete that data and force the Watch to start taking reading every 10 minutes again?
I've followed a lot of this thread but not all of it, so hopefully this doesn't sound stupid. I can't fully understand their reasoning here, but maybe after the watch has enough data from accelerometer AND heart rate it decides it can get away with accelerometer only (saving battery)? With occasional extra readings to keep up with increases/decreases in fitness?
Maybe Apple is satisfied that they know your moving heart rate based on accelerometer and have no need to measure it. But they are interested in the resting heart rate as that is a great measure of fitness.
Would unpairing/repairing the Watch delete that data and force the Watch to start taking reading every 10 minutes again?
Stupid change. Were there really enough failed attempts at measuring the heart rate that they had to effectively can this feature? For me, I found that it measured accurately no matter what I was doing. And as someone who spends a fair bit of the day on the go, my watch now only measures my heart rate about four times a day. It may as well not bother.
But isn't resting heart rate the best measure of health? Very active heart rate would be measured by the exercise app. And accelerometer data and known stride etc should mean that in-motion measurement is pretty good even without heart rate. By using your resting heart rate it still should have a good idea of your heart rate while in motion.
----------
All the engineer investigating and stuff happening with people from Tim Cook's office is pretty dodgy as it highlights bad communication within the company. But it does seem as though this change is intentional and Apple thinks it knows what is going on without needing to measure every 10 minutes while you're moving.
Apparently it is the best measure of health. But I'd like to know what it goes up to during the day when I'm playing the piano, especially during difficult passages. OS 1.0 did this and I didn't see any ill effects. Maybe Apple needs to provide a metric in the health app for resting heart rate, and one for active heart rate.
I'm trying to sort all this out to decide how important this change is - at least for me.
In any case, it was just wrong to make a change like this, with no info in the update about it and still having the old info about tracking HR on the website!
My first feeling was, "Why on earth do you STOP taking samples when you are moving?" Certainly seemed counter-intuitive! If it was because it would be more inaccurate - even if just moving your arm - then wouldn't this be an issue during exercise? Maybe not if, during exercise, a different sensor is used. I'm still unsure about this.
So I tried to figure out if there was some pattern to the decision. That is, the watch is really just going to measure two things - true resting HR every 10 minutes (assuming you don't move at the wrong time!) and HR during more serious exercise. But completely leave out anything in between, assuming you don't set it for an activity even though you aren't doing anything special.
With this in mind, I took my fully charged watch at 8:30 AM and turn on an activity ("Other"). Let it run for 2 hours. For most of this time I was sitting. Occasionally getting up, but very little walking around the room. As expected, it pretty much gave a HR every 5 seconds.
But there is an obvious problem with this approach since it used 20% battery in just 2 hours. Normally when I check, with no activity active, it would only lose about 2% an hour so this battery depletion was 5 times faster! Obviously not good for full-time use since it would be zero after 10 hours with luck. It also seemed my phone got warmer.
I'm not sure how fast it drained before the update, but I was fine with it.
After the two hours, I stopped the activity. Don't know why, but in the 3 hours after that, the battery dropped 9% (3/hour). Since that was still worse than any other test I've done, I don't understand that at all.
Since turning off the activity, I have gotten 8 HR checks in almost 4 hours.
One thing that someone might want to test, if you are patient - start with assuming the rate of checking once every 10 minutes would normally happen right when the minute changes.
Next, once you have a check show up on the Health app, about 10 seconds before the 10 minutes is up, move your arm around for 20 seconds (thus 10 seconds before to 10 seconds after the time it should check). Repeat this for each 10 minute time to check.
This is to see if you skip HR checks, even when resting, simply because at a particular time you happened to move your arm and thus cancel the check. It may account for why I can be sitting a long time and usually get 10 minutes checks, but occasionally it will skip one.
Of course, the 10 minute check may not necessarily occur when the minute changes. If not, more testing would be needed.
I may try this myself, but like I said, it will demand patience - and an ability not to miss the time to move my arm.
Oh, one more thing. The reason I want it checking at least every 10 minutes is because I've had some health issues which no one seems able to figure out, but there could be a heart factor involved. So I thought, if some episode occurred, it would be good to be able to look at the HR for when it did and see if it had an unusual change in speed. Not a common need, but it could be helpful.
So maybe Apple can give people an option for this. Let them set how often, no matter what, they want the HR checked no matter what when not doing an activity. Personally, I'd let it go as low as at least 5 minutes, but as high as you want. I am perfectly capable of figuring out how much battery life I need.
In any case, it was just wrong to make a change like this, with no info in the update about it and still having the old info about tracking HR on the website!
My first feeling was, "Why on earth do you STOP taking samples when you are moving?" Certainly seemed counter-intuitive! If it was because it would be more inaccurate - even if just moving your arm - then wouldn't this be an issue during exercise? Maybe not if, during exercise, a different sensor is used. I'm still unsure about this.
So I tried to figure out if there was some pattern to the decision. That is, the watch is really just going to measure two things - true resting HR every 10 minutes (assuming you don't move at the wrong time!) and HR during more serious exercise. But completely leave out anything in between, assuming you don't set it for an activity even though you aren't doing anything special.
With this in mind, I took my fully charged watch at 8:30 AM and turn on an activity ("Other"). Let it run for 2 hours. For most of this time I was sitting. Occasionally getting up, but very little walking around the room. As expected, it pretty much gave a HR every 5 seconds.
But there is an obvious problem with this approach since it used 20% battery in just 2 hours. Normally when I check, with no activity active, it would only lose about 2% an hour so this battery depletion was 5 times faster! Obviously not good for full-time use since it would be zero after 10 hours with luck. It also seemed my phone got warmer.
I'm not sure how fast it drained before the update, but I was fine with it.
After the two hours, I stopped the activity. Don't know why, but in the 3 hours after that, the battery dropped 9% (3/hour). Since that was still worse than any other test I've done, I don't understand that at all.
Since turning off the activity, I have gotten 8 HR checks in almost 4 hours.
One thing that someone might want to test, if you are patient - start with assuming the rate of checking once every 10 minutes would normally happen right when the minute changes.
Next, once you have a check show up on the Health app, about 10 seconds before the 10 minutes is up, move your arm around for 20 seconds (thus 10 seconds before to 10 seconds after the time it should check). Repeat this for each 10 minute time to check.
This is to see if you skip HR checks, even when resting, simply because at a particular time you happened to move your arm and thus cancel the check. It may account for why I can be sitting a long time and usually get 10 minutes checks, but occasionally it will skip one.
Of course, the 10 minute check may not necessarily occur when the minute changes. If not, more testing would be needed.
I may try this myself, but like I said, it will demand patience - and an ability not to miss the time to move my arm.
Oh, one more thing. The reason I want it checking at least every 10 minutes is because I've had some health issues which no one seems able to figure out, but there could be a heart factor involved. So I thought, if some episode occurred, it would be good to be able to look at the HR for when it did and see if it had an unusual change in speed. Not a common need, but it could be helpful.
So maybe Apple can give people an option for this. Let them set how often, no matter what, they want the HR checked no matter what when not doing an activity. Personally, I'd let it go as low as at least 5 minutes, but as high as you want. I am perfectly capable of figuring out how much battery life I need.
Agreed. I totally understand people being disappointed because they've given us something then taken it away and acted like it was a bug instead of clearly explaining the change. Apple likes to simplify things and dumb them down for the masses. But people who closely monitor heart rates (serious athletes especially) know what they want and don't want to be treated like idiots.
Best post of the day ^^^^^
Exactly correct. I am 51, and have been a competitive runner / cyclist / triathlete since high school and have used owned/used just about every piece of personal sports tech known to man at one point or another. So I am very familiar with using technology to hone my performance.
It is insulting for Apple to make these kind of changes under the cover of darkness with the hope no would care or worse understand the significance of this obvious regression of features.
http://9to5mac.com/2015/05/30/apple-says-watch-os-1-0-1-attempts-to-record-heart-rate-every-ten-minutes-but-wont-if-arm-is-moving/
This is absolute ********. How am I suppose to get an accurate look at my pulse for the day if you only take my pulse when I'm not moving. We all know if your not moving your pulse is going to be lower. Really wtf
.
Let's ALL email Tim about this. The only way to change things is to have tons of people email Tim on the same topic.
----------
Really what benefit does taking my pulse only when I'm not moving give me. All it gives me is biased heart rate results when looking at the day as well as less results.
This is absolute ********. How am I suppose to get an accurate look at my pulse for the day if you only take my pulse when I'm not moving. We all know if your not moving your pulse is going to be lower. Really wtf
. Let's ALL email Tim about this. The only way to change things is to have tons of people email Tim on the same topic.
----------
For me it's more about the right and wrong way to do it then being use to the old way.
Really what benefit does taking my pulse only when I'm not moving give me. All it gives me is biased heart rate results when looking at the day as well as less results.
I did some testing - not a lot because it takes a lot of time when dealing with 10 minute gaps - but this is what I found. Maybe others can test some more to confirm.
I worked on the assumption that, when not moving, the HR will be checked at the BEGINNING of the 10th minute. That is, if the watch stored a reading at 4:00, then the next check will be right when the minute changes to 4:10. My test was to confirm that. All tests were when just lying in bed.
My first test, I just casually lifted my arm, up and down, starting 10 seconds before the minute changed to 10 seconds after. Apparently this wasn't enough movement to block the check because that 10 minute check did register with the Health app. So the next time, doing about the same time range I moved my arm up and down much more vigorously. Sure enough, I never got a reading for that 10 minute mark in the Health app.
From that point on I mostly alternated between not moving - to make sure I would still get a reading - and moving my arm a lot. However, I kept reducing the time range. My last test I only started moving my arm one second before the minute mark and ended one second after. Sure enough, I still didn't get a reading.
So, it seems if you just happen to be moving your arm around (how much is still open for more testing if anyone wants to try) for even two seconds - maybe less - you'll miss that 10 minute HR check. I don't think it matters how much you move around during the rest of the 10 minute period. But it might be worth testing to see what happens if you are pretty active for almost the whole 10 minutes but stop a minute - or seconds - prior to the minute change just to see if it registers a heart beat or not in that situation.
I worked on the assumption that, when not moving, the HR will be checked at the BEGINNING of the 10th minute. That is, if the watch stored a reading at 4:00, then the next check will be right when the minute changes to 4:10. My test was to confirm that. All tests were when just lying in bed.
My first test, I just casually lifted my arm, up and down, starting 10 seconds before the minute changed to 10 seconds after. Apparently this wasn't enough movement to block the check because that 10 minute check did register with the Health app. So the next time, doing about the same time range I moved my arm up and down much more vigorously. Sure enough, I never got a reading for that 10 minute mark in the Health app.
From that point on I mostly alternated between not moving - to make sure I would still get a reading - and moving my arm a lot. However, I kept reducing the time range. My last test I only started moving my arm one second before the minute mark and ended one second after. Sure enough, I still didn't get a reading.
So, it seems if you just happen to be moving your arm around (how much is still open for more testing if anyone wants to try) for even two seconds - maybe less - you'll miss that 10 minute HR check. I don't think it matters how much you move around during the rest of the 10 minute period. But it might be worth testing to see what happens if you are pretty active for almost the whole 10 minutes but stop a minute - or seconds - prior to the minute change just to see if it registers a heart beat or not in that situation.
Presumably, calibrating the watch is enough to figure out a good correlation between your heart rate, level of activity, pace and calorie burn. While I would prefer to have it back to the old system (every 10 mins without fail), I haven't experienced any loss of my move/exercise/stand rings because of the change. If anything, they've become more accurate, so I can see the reasoning in that regard.
Bare in ind, I have calibrated the watch just as they advise in their user guide (outdoor walk/run using workout app for 20 minutes with your phone in your hand for a good gps reading).
Last edited:
Great info. It's a pity Apple only describe it in the vaguest terms so thank you for doing this
Funny thing, all the stories about the apple Watch said that Apple intended to have a number of health sensors but then decided to remove them from the first release due to innacuracies.
Now they say their IR HR sensors are not accurate enough to measure HR when you are moving. Then what's the point!!!
Release a damn watch with no sensors, that only shows the time.
I was pretty happy with the way the infrared sensor worked. Obviously for me the active (green) data will reflect activity better, but it was very nice to have the other data even if a little innacurate.
I could, however, vouch for its accuracy. With v.1.0 I could go back to the HR data and identify exactly, when I was at meetings, and when, within that meeting, my HR increased due to me having to speak up or me presenting data t external partners.
The thing is, this is so obviously bull crap.
Now they say.. they measure every 10 minutes, except if your arm is moving.
What is the point of that rule? You control the HR, you control the accelerometers. If that was your intended behaviour, you would do it THE OTHER WAY AROUND. Measure movement and any time YOU ARE NOT MOVING, then kick in the HR monitor sensor and measure resting rate.
The fact it's done the other way around only evidences this is a desperate patch to fix something. And that something is, most likely, battery life. In my opinion.
Now they say their IR HR sensors are not accurate enough to measure HR when you are moving. Then what's the point!!!
Release a damn watch with no sensors, that only shows the time.
I was pretty happy with the way the infrared sensor worked. Obviously for me the active (green) data will reflect activity better, but it was very nice to have the other data even if a little innacurate.
I could, however, vouch for its accuracy. With v.1.0 I could go back to the HR data and identify exactly, when I was at meetings, and when, within that meeting, my HR increased due to me having to speak up or me presenting data t external partners.
The thing is, this is so obviously bull crap.
Now they say.. they measure every 10 minutes, except if your arm is moving.
What is the point of that rule? You control the HR, you control the accelerometers. If that was your intended behaviour, you would do it THE OTHER WAY AROUND. Measure movement and any time YOU ARE NOT MOVING, then kick in the HR monitor sensor and measure resting rate.
The fact it's done the other way around only evidences this is a desperate patch to fix something. And that something is, most likely, battery life. In my opinion.
I did a test last night on running an activity full time. I chose "Other" for the activity and ran it all night so, other than occasional checking of the time or battery level any time I woke up, that was all it was doing.
This was on a 42 mm watch. After 9 hours it was at 10% (I started at 100%) so it lost 10% per hour. This matches a test I did earlier in the day for just 2 hours.
Because of my schedule, I could probably run it this way all the time since I can recharge easily in the morning, around lunch, and around supper time. But obviously most would find that hard to do.
Also, I do not know if other activities will make a difference. For example, cycling outdoors would certainly be tied in with the iPhone GPS - does this use more battery on the watch compared to "Indoor cycling" where no GPS would be needed? And I don't even know if "Other" involves the GPS or not so more testing is needed - unless someone already knows and can post.
Then there is the question of how much difference any of these things make on the iphone battery - certainly GPS would have a big impact.
Apple really needs to provide an explanation. Maybe they have an argument that makes sense that we haven't figured out, but so far it seems you simply have a choice of a HR when completely resting or when doing an activity (not counting using the Glance - I'm more concerned about things it does pretty much automatically, not something I have to constantly check manually).
It would be useful to know the battery drain of the 38 mm watch for comparison too.
In my normal usage - which is pretty light - it usually takes 2% per hour so the activity is at least 8% more each hour, checking every 5 seconds for the HR. Some basic math (assuming I'm not missing something big) would show that the 720 checks/hour costing you that 8% would mean that each check takes 0.0111%.
So, if we only check once every 10 minutes, that's just 6 total checks costing a total of 0.0666% per hour. Which would mean a check every 5 minutes would take 0.1332% in an hour. Heck, one every minute would take just 0.6666 per hour. Translate that over the claimed 18 hour battery life and the HR check would only total about 12% if checked every minute.
All these numbers strike me as pretty darn low - at least on a 42 mm watch - so maybe I'm missing something. But if I'm not, it sure doesn't seem like they are making a meaningful impact on battery life with this change so it must be something else that hasn't been explained yet.
Or a really bad excuse for a bug!
Put in an option to let the user set the normal interval for checking, from every minute to as little as once every hour or something.
Another improvement they could make, though not ideal, would be to deal differently with the "moving arm" issue. It's just not right that moving your arm for 2 seconds (or less) at the wrong time may cause you to completely miss a reading. The watch is already clearly aware of whether you are moving around or even just moving your arm since it has to know that to NOT take a HR sample. Assuming the real reason for this change is because the movement gives bad readings with the red LEDs, why not, if the watch is moving at the 10 minute mark, not check HR but as soon as you stop moving, go ahead and check, regardless of the time. So you might have something like this as a result:
53 12:00 PM (first normal HR check)
54 12:10 PM
52 12:20 PM
55 12:32 PM (delay of 2 minutes because of movement at 12:30)
The next check could be handled one of two ways. Either check again at 12:40 or restart the 10 minute interval from 12:32 so the next check would be 12:42.
Seems to me that if it isn't battery, but just accuracy because of movement, either of these approaches would solve that issue and give us pretty much what we had before - except for those cases where you are moving all the time.
I'll have to look at this one soon.. what amazes me though is that ppl are selling this Basis Peak on ebay for 189-299$ when it can be bought new from amazon for 180$? What is that?
Exactly, the last part of your message makes sense.
To me, this is a last minute bug that they patched with a half baked solution.
So now they say the 1st trigger is time (10 minute) and second trigger is lack of motion.
That doesn't make any sense, exactly for the reason you mention. You can be still for 9min 55 sec that if you move in the last 5 seconds, that movement will stop the HR reading until 10 minutes later.
A logical way would be to have Movement as 1st trigger, time as 2nd trigger. So every time the watch detects you aren't moving, check how long ago was your last reading, if time is > 10 min, then trigger a HR reading.
This goes to prove that this is not actually intentional behaviour but just some last minute patch.
To me, this is a last minute bug that they patched with a half baked solution.
So now they say the 1st trigger is time (10 minute) and second trigger is lack of motion.
That doesn't make any sense, exactly for the reason you mention. You can be still for 9min 55 sec that if you move in the last 5 seconds, that movement will stop the HR reading until 10 minutes later.
A logical way would be to have Movement as 1st trigger, time as 2nd trigger. So every time the watch detects you aren't moving, check how long ago was your last reading, if time is > 10 min, then trigger a HR reading.
This goes to prove that this is not actually intentional behaviour but just some last minute patch.
Hard to say for sure which is the best approach. If movement is first, you are constantly checking for movement until a time check qualifies - then go right back to constant movement checks. Would certainly work, but could affect the battery more.
If you check time first (and obviously the watch has to keep the time anyway), you avoid any battery drain from measuring movement for 10 minutes at least. At that point, once it shows no movement, it would skip 10 more minutes.
Of course, I have no idea how much battery may be affected by these things, but I think the time check first at least has potential to reduce battery drain.
Regardless of the approach taken, it sure looks like something that should be easy to fix.
I don't see how people can say Apple made this change for accuracy. Workouts take your pulse the entire time and you're likely moving your arm the entire time.