Heart Rate Extremely Inaccurate, Microsoft Says Band is Entry-Level Device

[SteveVII]

I think you need to change the title of your post to "for you its inaccurate". I just went for a run this morning with my motoactv watch/garmin chest strap hr monitor and the microsoft band.

Overall they both matched up with min/max and average heart rate for the complete 40 minutes. I even waitied 10 minutes after the run for the cool down and they both were in sync.

now when i was running were there times one read different than the other? Sure but ones on my wrist and ones on my heart within some time when i looked down they were in sync.

I am very happy with the readout as I was eager to test after reading this post last night.

If anything the GPS took a while to capture under some cloud cover that is my only gripe.

Check it for yourself guys before believing someone trying to make a claim for all of us.

Sanva, if you read my posts, I specifically asked about everyone else's results to determine if it is normal or not. I shared my experience, and asked others to chime in to help determine whether my results were normal or not. Unfortunately, the title only allows for so many characters.

I'm happy to hear that it was accurate for you. I experienced similar results initially. I started noticed the discrepancies when doing non repetitive motions (weight training) and going on the elliptical for cardio (maybe the bouncing was screwing it up). If you use it primarily for running and it works great, I'm happy for you. As for me, and others who were looking for this to be a fitness-first device for various activities, it is failing to deliver so far. I take my workouts very seriously, and when it comes to cardio, a difference of 10 BPMs makes all the difference regarding which "zone" I need to be in.

As I mentioned, I want to like this... I was at the Microsoft store 45 minutes early to make sure I got one. Unfortunately, many of the key advantages that seemed promising in my opinion (24 hour continuous heart rate monitoring where it is actually 24 hour heart rate sampling, accurate heart rate during non-repetitive motions) don't seem to be built into the device. That being said, I again see the value in what Microsoft released, however, the press release/marketing was definitely misleading.

Also guys, any idea if the other sensors (skin temperature, Capacitive sensor, Galvanic skin response, etc.) are being used yet? I haven't found any indication that this is being tracked or used as calorie burn seems to be entirely based on steps and heart rate. This is something that may make me want to keep this as an "all day fitness tracker", but so far the metrics are nowhere to be seen. Also, sleep tracking seems to spit out a lot more data points for heart rate throughout the night vs. viewing all day. I wonder if the band is tracking this and the Microsoft Health app is only programmed to show hourly averages?

 

[SteveVII]

If you guys allow me to summarize my tl;dr post, I think the HR measuring alogithm of the band is set to ignore rapid HR changes, and it appears to have been set too aggressively. That means if your real heartrate drops rapidly, the band will show 'out of sync' and hold the last high value. See the graph below especially near the end, band in purple, garmin FR305 in red. If this is the entire problem, it seems fixable.

If this theory is true, then people doing exercise with constant effort will report good results and rarely lose sync, and people doing interval training etc will report bad results.
http://forums.windowscentral.com/attachments/microsoft-band/86308d1414905852-hrmtest.jpg


Tip for the MS coders: if the motion data suddenly calms down AND the heart rate suddenly drops: believe it. If the motion stays the same, but the heart rate changes a lot: ignore it. Not fully fool-proof, but it might help.

Joe, which mode was your band on with these results? I wonder if training mode tries to look for these intervals, whereas running mode assumes it will be more constant. I agree with your analysis here. During my weight training sessions, my heart rate never dropped below 100 during resting. It was like the band was assuming that I should be over 100 rather than taking my actual HR.

 

[debad]

I did test it for myself and it was way off, for tennis at least. Before I give up on it, I'll try a plain run to see if I get good results like you did.

During your run, was the HR locked with the solid heart the whole time? Which side of wrist did you wear the display?

The TomTom in the review seemed super accurate. I think the optical sensor can work, but my Band is touchy and the sampling rate for readings is too low. Before I even took it outside, waving my arm around caused the solid heart to change to the outline evrytime. The fact that the hr number changes even when it's the outline heart makes me think it is relying on a projection much of the time as opposed to real time readings. Moment to moment, mine jumps around a lot, frustratingly so.Probably it does best with a more constant heart rate pattern, like a nice steady run. Again, the sensor is pretty tight on top of wrist (display inside of wrist), not a lot of hair. Doesnt seem like light's getting in, but could be.

 

[mhc48]

If you guys allow me to summarize my tl;dr post, I think the HR measuring alogithm of the band is set to ignore rapid HR changes, and it appears to have been set too aggressively. That means if your real heartrate drops rapidly, the band will show 'out of sync' and hold the last high value. See the graph below especially near the end, band in purple, garmin FR305 in red. If this is the entire problem, it seems fixable.

If this theory is true, then people doing exercise with constant effort will report good results and rarely lose sync, and people doing interval training etc will report bad results.
http://forums.windowscentral.com/attachments/microsoft-band/86308d1414905852-hrmtest.jpg

Tip for the MS coders: if the motion data suddenly calms down AND the heart rate suddenly drops: believe it. If the motion stays the same, but the heart rate changes a lot: ignore it. Not fully fool-proof, but it might help.

That's specifically what I was trying to test this morning and I didn't find it to be so. Granted I cannot compare what I saw to another device's readings, but as said before, I can certainly correlate what I saw and later read on the chart to my own actions.

BTW, I'm not trying to dispute or knock your findings or theory, just to add to the available data.

 

[Joe920]

BTW, I'm not trying to dispute or knock your findings or theory, just to add to the available data.

No problem, I am just happy that I found a theory that matches my data and that explains why adjusting the strap doesn't do anything. I'm hoping we'll see more overlaid curves so we can see if anything like that is going on.

 

[Daniel Rubino]

i have no idea how the algorithm and the light sensor interact, but this seems like it would clearly be a less reliable method than the medical standards.

Long story short: optical is less accurate than electrical but more practical for everyday use.

Pulse oximeters used to determine blood oxygen levels also use light and a photodetector. The tech can also pick up pulse rate. Having said that, in a medical situation, we always used ECGs for patients due to their accuracy, but even then, they are highly dependent on placement and body fat composition.

The fact is, even chest straps are going to be less accurate than a full 12-lead ECG. Often enough, you need at least 2 contact points for a semi-accurate ECG in a hospital, but 12-lead ECGs are still the best. The same is true with EEGs for neuro-- there are different channel setups, all with increasing levels of accuracy. In fact, software often infers proper source localization of EEGs signals these days instead of raw meaurements.

The takeaway: even chest strap HR monitors are only so good, but there is a very good reason why they are not used in hospitals either

The technology we have in the Band should be considered semi-accurate, but not enough for a full medical diagnosis. This is all trickle-down medical technology, it's getting better, but not nearly as good as the 'pro' stuff. It's no different than how crappy cell phones were in the 90's compared to landlines.

I have two different chest straps (Runtastic, Adidas) that I'll be comparing to the Band later on, though I would be shocked if the Band was more accurate.

It's trade-offs. The Band brings this tech down to an affordable consumer level, and it is more convenient than a chest strap (which can be a PITA to use). But even the coveted chest straps are not nearly as good as medical-grade ECG setups, let's not kid ourselves.

 

[Joe920]

Joe, which mode was your band on with these results? I wonder if training mode tries to look for these intervals, whereas running mode assumes it will be more constant. I agree with your analysis here. During my weight training sessions, my heart rate never dropped below 100 during resting. It was like the band was assuming that I should be over 100 rather than taking my actual HR.

I had it on exercise mode, not running mode. Beginners mistake.. For biking I should have used 'run' mode and activated GPS manually.

 

[SteveVII]

Thanks, Daniel. Interested to hear your results. I think it's pretty clear that wrist HR monitors are not especially accurate. However, based on Microsoft's press, it seemed like this should have been a step in the right direction. Instead, it doesn't seem like the optical sensor has as good of accuracy as many older devices. Let us know what you find.

 

[Joe920]

I have two different chest straps (Runtastic, Adidas) that I'll be comparing to the Band later on, though I would be shocked if the Band were more accuarate.

Cool! First thing: in the article please push Runtastic.com to integrate with MS health data, because as of now it doesn't. And second, I hope you'll be able to overlay the HR curves, that makes for a much easier comparison of accuracy. Looking forward to the review!

 

[surfacedude]

. But even the coveted chest straps are not nearly as good as medical-grade ECG setups, let's not kid ourselves.

i agree and think this is an important. and just to be clear, i didn't say chest straps were as accurate ekgs. i said that as far as i know, they tend to be more accurate than (consumer grade) light sensors.

i think these these consumer health devices should not be confused for medical devices. the same goes for home blood pressure monitors. they may help you keep track of blood pressure, but no doc will make a diagnosis with one.

we have to have realistic expectations, that's my point, but my earlier post may not have clearly conveyed that.

 

[Joe920]

I wonder if the thread title should be edited to state "MS phone rep says ..". I mean the title is long enough as it is, but right now it reads as if this is some kind of official MS stand. Admittedly it is kind of an entry level device, but I'd be surprised if this were the official message.

 

[gpobernardo]

I once* had the Instant Heart Rate app and it was quite accurate... most of the time**. It requires covering the LED flash beside the main camera and the entire main camera lens in order to detect discolorations underneath the skin caused by oxygenated blood flow - this makes prolonged continuous use impractical, unless you're willing to drain your battery in less than an hour. So I was all ears when the Microsoft Band was launched, and I was especially expecting that the band performed as well as the app but with a much longer battery life.

However, it's news and feedback like these that reflect the insights of those who already have had their chance to interact with the product; seems like I'll have to stick to wearing automatics for now (as I have in the past 18 years).

*- Doesn't work with the Lumia 1020 since the LED flash is too far away from the main camera lens cover. I used the app with the HTC 8s.

**- I'm saying, "most of the time," because it occasionally "failed" to detect any pulse (usually because my finger was not covering the entire lens or because there were interfering light sources nearby), but when it did it never missed a beat. I'm not sure how the band works since I don't have one yet, but I'm guessing that the inaccuracies in the Microsoft Band are caused by the HRM not being in total and continuous contact with the skin of the wearer, causing either false or phantom pulses to be detected. The HRM could be very accurate if it remains in contact with the skin and if it runs continuously, but that still doesn't make it as reliable as professional medical equipment.

 

[greyskytheory]

If anyone is counting on this device to track your workout metrics in deep, accurate detail you're looking in the wrong place. You'll need to get a Garmin, Suunto etc. For all of the functionality the Band delivers it is a great device but I didn't expect it to be dead on accurate or provide tons of insight into my workouts. We all need to manage our expectations. This $200 wearable does things my $500 Suunto can't. happy it provides a depiction of my daily activities.

 

[SteveVII]

If anyone is counting on this device to track your workout metrics in deep, accurate detail you're looking in the wrong place. You'll need to get a Garmin, Suunto etc. For all of the functionality the Band delivers it is a great device but I didn't expect it to be dead on accurate or provide tons of insight into my workouts. We all need to manage our expectations. This $200 wearable does things my $500 Suunto can't. happy it provides a depiction of my daily activities.

I think it is becoming clear to most people that this is the case and we are coming to accept it. I think at this point, a handful of us are disappointed that this isn't a "game changer", which is how the press responded to it. I think if we shouldn't count on this device to track my workouts accurately, then Microsoft should take down this post asap: Microsoft Band, the first wearable powered by Microsoft Health, keeps fitness and productivity insights a glance away | News Center .

 

[Joe920]

I think it is becoming clear to most people that this is the case and we are coming to accept it.

Maybe in part, but some of us want to see more data! I don't expect the device to outperform high end devices, but it kind of looks like it could do pretty well. Out of curiosity, does anyone know how the raw data of this kind of sensor looks? Maybe an enterprising user can open this sensor up and hook up the detector readout to a good ole' digitizing oscilloscope? Might be hard to get to the contacts though.

 

[debad]

Will the iWatch be more advanced? Talk about that made me think MS might have made a significant step in wrist monitoring. Should have waited till the reviews were out, but was fun testing anyway.

 

[phillyphan23]

Which way are you wearing the band? I normally wear it "reversed" (display on palm side of hand). Used it during a stationary bike session and found the band was reading much lower than the bike heart rate display with Polar chest strap sensor. I switched the band halfway through the 30 min session and both the band and bike tracked pretty closely.

 

[ChumsFuture]

I just climbed up and down stairs with either a walk or jog in between each set of stairs.

I have a theory. I believe there's a severe lag with the heart rate displayed on the Band compared to actual heart rate. I noticed when I'm walking after finishing going up a set of stairs, my heart rate displayed was high, yet when I'm actually on the stairs going up, huffing and puffing, the displayed heart rate is more like my actual walking heart rate. Yet, after my workout, I look at the app and I've got all the proper data, with all it's peaks and valleys, that are consistent with this type of workout.

If your workout is steady like a long jog, the displayed heart rate should be close to actual. However, workouts with peaks and valleys in heart rate, the displayed heart rate is lagging and playing catch up.

 

[Joe920]

So for anyone doing tests, please note:
- whether you wear the band facing in or out (inside of wrist or on top of wrist)
- which mode you use (running or 'workout')
- which type of sport you're doing
- which comparison device you're using if any (brand, HR strap?)
- whether your band shows 'out of sync' (heart icon outline only) or 'in sync' (solid heart icon)

Personally I see pretty good agreement when the band claims to be in HR sync, the problem being that it loses sync a lot.

 

[shawncade]

You are correct, the Band's heart rate monitor is very inaccurate and erratic, to the point it is useless. I have compared it to a Polar that has proven accurate during a stress test and when actually locked the rate can be 40 bpm's off. Also, it is always searching for a heart rate.

If you car about an accurate HR don't buy the band. I will be returning it but I really wanted to keep it.