Hume Band 2.0 - Full Scientific Review!
In a Nutshell
The U Band 2.0 delivers unreliable heart rate tracking during exercise with frequent artifacts, half-rate measurements, and low correlations (0.34-0.42) against reference devices. Sleep stage tracking performs poorly at 44% deep, 53% light, and 32% REM agreement with EEG validation, ranking it among the worst devices tested. Nighttime heart rate shows a suspicious hard-coded 45 bpm minimum that artificially doubles readings when actual heart rate drops lower, making the device unsuitable for serious use.
These notes were generated by AI and may contain inaccuracies.
A few weeks ago, the first look at the You-Man 2.0 was published, but this video is not a repeat of that video. The original video was a relatively quick introduction based on the data available inside the You-Map that could be synced to Apple Health. This video goes much further because shortly after publishing it, something completely unexpected arrived in the inbox - the GDPR data export containing, in principle, all the data that You-Man has stored. The surprising part is that the export had been requested 5 months earlier for the You-Man 1.0 review. This export should theoretically give considerably more data and allow testing the You-Man 2.0 in much greater detail than could be done before. However, there are still some significant limitations. You-Man uses some rather unusual data formats and the level of detail they retain seems somewhat randomly variable between measurements and periods. Nevertheless, this is likely the most complete and in-depth analysis of You-Man 2.0 data that will be found. You-Man does not exactly have the best reputation. That reputation has been justified, particularly because the company previously used what appeared to be fake scientists in their marketing, or at least their scientists were also stock image models. However, with the You-Man 2.0, the company does seem to be trying to improve. Fake scientists could no longer be found on its website, and the band itself may represent a step in a better direction. Of course, good marketing is not the same thing as good science. The question is, what does the data from the You-Man actually tell us?
The reviewer is a post-doctoral scientist specializing in biological data analysis, and that will be the focus of this video - data and data analysis. Using both the GDPR export and reference devices, a close look will be taken at what the U Band 2.0 measures, how accurate those measurements appear to be, and what information U actually retains in the end, at least the data that was received. The goal is to go far beyond the original introduction, and by the end of this video, have a clear and realistic understanding of what the U Band 2.0 can and cannot do, so exactly what is being bought is known.
For spinning, one example session is shown with clock time along the horizontal axis and heart rate along the vertical axis. The reference device is the Polar H10 ECG chest strap, one of the most reliable ways to measure heart rate, shown in blue green, and the U Band measurements that could be accessed are shown in red. Several intervals were done, and the measurements taken by the U Band weren't half bad, but it just recorded very infrequently, or at least stored it very infrequently. A few measurements are available, and most of those align rather well with the measurement of the Polar H10 for this spinning session.
The second spinning example session had more trouble, so it sometimes detected a too low heart rate. In all these moments, it really struggled a little bit more. The third session looks quite okay again, but then the fourth example spinning session detected both a too high heart rate and too low heart rate. So honestly not that promising.
There wasn't enough data to make the normal full overview for spinning, so it didn't pass the filtering process, but where it did meet the filtering process was for running. The heart rate according to the reference device is on the horizontal axis and the Uman 2.0 along the vertical axis, where each dot is a single heart rate measurement. If they would perfectly agree, all points should be on or close to the blue line. They are in some cases, but in other cases, it seems to be measuring half the actual heart rate. That also leads to a low correlation of 0.42, so really not that good. The maximum correlation is one, so 0.42 is not really close to that. This is some of the worst performance seen.
Looking at individual running sessions, it's not always bad. One running session is sometimes quite close, but sometimes detects a too high heart rate. Another session sometimes has a too high heart rate. One session had weird data - every fourth or fifth measurement, there's all of a sudden a peak in heart rate. Not sure what's going on there, but otherwise it detected about half the actual heart rate. Another running session looks a bit better with still too low and too high measurements - near the end it's too low and in some moments it's too high. This just doesn't look very promising compared to the competition, like for instance the $100 Fitbit Air. It's also a lot less transparent in how much data it's actually gathering and storing, and access to the data is super limited.
For biking outside, a similar overview is shown with the reference along the horizontal axis and the Uman 2.0 along the vertical axis. Again, there's a super low correlation at 0.34. Yes, some points are along the blue line, but many times the Uman 2.0 detected a way too low heart rate, so not very good. Looking at individual sessions, quite often a too low heart rate is seen with some moments where it's actually on point. In one longer session, it was able to detect heart rate sometimes, but also quite often it detected too low heart rate. So really not that trustworthy.
The most interesting results are for weight lifting. The reason these are most interesting is because here for some reason much higher resolution data was accessible, which is why there's many more points in this plot. Again, the reference is along the horizontal and the U-Band along the vertical axis, where those points are wanted to be close to the blue line. Especially in the higher heart rate range, the U-Band detects too low heart rate, indicating that each time a set of exercises is done there's an increase of tension, and the U-Band just cannot deal with that increased tension.
Looking at individual sessions, for some reason one session still had low resolution data, but it only detected lower heart rates. For the high resolution ones, it picked up on none of the peaks of heart rate, and often when there was an increase it actually detected a decrease in heart rate. Another session only detects the baseline heart rate, but not the peaks in heart rate.
Given that only low resolution data is available for some exercises, an official score wasn't calculated, but the feeling is that this is quite bad heart rate tracking overall. Just not trustworthy enough to use, because there are enough artifacts and weird things going on that the heart rate tracking data cannot be trusted.
Sleep stage tracking is tested at night, when lying still and there's a lot less interference from movement and also light. It might measure metrics correctly and also estimate sleep stages. The sleep stage tracking performance is compared against the Zeemee EEG headband, which is an EEG headband that can measure brain waves and was specifically designed for sleep studies. It's sort of a silver standard, so good enough to test different devices. The reference device is on top and on the left is the U band. If the two would perfectly agree, all values along the diagonal should be 100%. This is not 100%. Yes, within each of these columns the correct sleep stage has the highest value, but still overall an agreement of 44% at deep sleep, 53% at light sleep, and 32% at REM sleep. Just not very good compared to the competition.
Sleep stage tracking is very hard, but comparing to the competition, the U band just is not very good. It's all the way down in the performance chart. The average sensitivity is along the horizontal axis, the worst sensitivity along the vertical axis. That means the best devices are towards the top right, and the U Band 2.0 is all the way down here, the worst device together with the U Band 1.0. Probably the same algorithm is being used. They're really just the worst out there. The U Band cannot be recommended for sleep stage tracking.
In testing, there are five or four devices that perform best. The top devices are different Google and Fitbit devices. Pixel watches and Fitbits are doing really well, and a Fitbit Air can be gotten for 100 bucks. That is really recommended over the U Band. The Whoop Strap also performs well, and in addition the Oura Ring 4 does really well in testing. Though, the Oura Ring 5 actually did a little bit worse. More in-depth testing about that will be done over the long term as well. Apple Watches did really well. Finally, the Eight Sleep Pod is the favorite sleep improvement device. The Eight Sleep Pod has really good sleep stage tracking at least on this body, but that's not what it would be gotten for. It's not a wearable, it's something that goes around the mattress. Regular viewers will know about this, but it's the main sleep improvement device because it can cool and heat each side of the bed independently. It goes around the current mattress. There's also a blanket that can be bought with it that cools and heats. In summer, this has been a godsend. There's so much trouble sleeping in the heat, and this has just changed the game because when there's a partner, she can set her temperature independently, and a cool sleep is preferred. It's super expensive though, so only buy it if it can actually be afforded. But if the best discount possible is wanted and to try it out for a while, the affiliate link should be used.
What would not be recommended is the U Band. The representatives from U have contacted a lot to sponsor a video or do some kind of sponsorship, but independent testing is done, and the data is just shown. The most that will ever be done is an affiliate link, which will even be done for mediocre devices. The data will just be shown and the decision can be made. But this is so bad that it cannot be recommended to anyone with good conscience. An affiliate link will also not be accepted from them in this case. Really not happy with the way they're performing and not with their past behavior. Of course, things can improve, but marketing being the main selling point is not liked. The You Band just is not doing very well.
Different metrics like Cohen's kappa can also be calculated for the You Band. But even with Cohen's kappa, the You Band 2.0 is one of the worst devices out there. Though in this case, actually some old Huawei devices are even slightly worse. But in general, all the Huawei devices, Coros devices, and the You Band would not be recommended for sleep stage tracking. Coros is very good at GPS though and also pretty good at heart rate, just not at sleep stage tracking.
In addition to measuring sleep stages, the You Band can also measure heart rate and heart rate variability at night. Very weird results were also gotten for that. Trying to correlate the reference device for heart rate against the You Band along the vertical axis gives this weird cloud of points and no idea what's going on. Also interestingly, the You Band seems to never measure a heart rate below 45. Something weird is going on and what is thought is that they might have hard coded a limit of 45 bpm at night and then say, "Okay, if you measure that, it's probably actually double the heart rate." Because if there's a noisy signal, you never know if it's double or half the heart rate. So if it's below that, let's say 40 bpm, it likely goes to 80 bpm. And that's exactly where we are right here. So that could be a partial explanation.
Looking at some of those measurements, for instance in the beginning of the night, heart rate was above 45 bpm and then the moment it drops below 45 bpm, all of a sudden it starts detecting a way higher heart rate. And that's about double the heart rate. So, what is thought is that there's some weird artifact where there might be a hard-coded limit of not measuring heart rates below 45, which is why this measurement is again accurate. That's the suspicion, at least. Something mostly similar is seen where around 45 BPM, all of a sudden way too high heart rates start being detected. The same, maybe it's not exactly 45, is happening where all of a sudden the U men starts detecting way too high heart rates. That's just what happens for many nights, and that seems to be especially true for one night where heart rate was just tended to be very low, so mostly below 40 BPM, so around 37, let's say. So, if that's doubled, about a heart rate of 70-74 is expected, so somewhere in this range. And indeed, this is where the U men ends up. So just not a very reliable tracker. It might do better if heart rate does not fall below 45 BPM. But a lot of training is done, which is why at night heart rate drops a lot, and that's just not very good. This is the suspicion, there isn't definitive proof, but this is what was found.
For HRV, or heart rate variability, some weird results are also gotten. But here it's not so sure what kind of HRV U uses. Testing is always done against RMSSD, and maybe U uses something else, because comparing the U band against the reference, the U band always detects a way lower HRV. So, not sure what's going on here. This doesn't dare to be tested in detail, because it's not sure what they're actually doing there.
Overall, some weird things are going on with the U band, especially this 45 BPM heart rate limit at night, the pretty bad sleep stage tracking, and also not so great heart rate tracking during exercise. In general, this device cannot be recommended. Curious to hear about experiences - if there's a heart rate below 45 BPM at night and if it reports a way too high heart rate. Really interested to hear. Overall, just get another screenless wearable at this point and it doesn't really matter whether it's a Whoop strap and Amazfit Helios strap or maybe one Amazfit Helios strap Pro, the Google Fitbit Air or the new Garmin Circa which at this point hasn't been received yet. Hopefully soon other reviews will always be a bit quicker with Garmin devices. But maybe once this video is edited there will already be an initial review of the Garmin Circa. That's a device that's really exciting. But the Whoop band is just not a band that would be recommended. It also comes sort of drop shipped from China so extra import fees might even be faced at this point in Europe. Overall, just get any other device from a reputable brand and don't get the Whoop band. Don't sue - just showing the test results and findings.
If one of these screenless wearables is gotten, or wanting to improve sleep with the Eight Sleep Pod or wanting to buy a watch in the end, whether it be maybe an Apple Watch or a Pixel Watch or another device or anything at all on Amazon for that matter, even something as small as toilet paper or coffee beans, wanting to support the channel and at the same time get the best discount possible, there are a bunch of affiliate links. Given that this video is being watched on the Whoop band, the video on the Amazfit Helios strap or the video on the Eight Sleep Pod might actually be liked.
Keep The Quantified Scientist in your library
Save the videos and channels worth coming back to, and find them again in one place.
