Possible Fix for inaccurate heart rate

[SunsetRunner]

So I have been toying with the Sense and it's HR monitor the last few days and think I have found a fix 

It seems their is a bug in the software in terms of how it auto generates custom min and max heart rates that cause the sensor to bug out and lag hence inaccuracies in HR, especially exercise.

I have found the following steps improved the Sense greatly:

1. Open Fitbit app

2. Scroll down to the Heart Rate tile and tap it (the one with the heart that shows your heart rate)

3. Tap the gear in the upper right hand corner of the graph.

4. Tap Heart Health listed under the exercise heading 

5. Check the box that says "Custom Max Heart Rate" 

6. Set the Custom Max Heart Rate to 200

7. Check the box that says " Custom Heart Zone"

8. Set upper limit to 200

9. Ser lower limit to 40

10 tap done in upper right hand corner.

11. Return to main screen and sync Sense.

12. After sync is complete hold down side button for 8 seconds until you see fitbit logo to reboot device.

13. After device reboots sync Sense one more time with the app.

You should notice the HR monitor is much more responsive to changes even at rest. You may have to play with those numbers a bit, but they are not your actual Max rates, they are just numbers used to calculate your Max rates.

It's my experience the auto generated numbers do not work and cause the HR sensor to lag. This appears to fix the heart rate issues with my spinning exercises. 

I hope this information helps others who are having issues.

[JohnnyRow]

@SunsetRunner  What were your default numbers before resetting, so we can know what direction you changed them?

200 and 40 might expand the range for some, shrink the range for others.

[SunsetRunner]

@JohnnyRow  When I checked the "Custom Max Heart Rate Box" the number that appeared was 178 (Which makes sense as my max heart rate should be 220 - my age(42) which would equal 178. However, It doesn't appear these auto generated values are used, or the actual hardware in the sensor itself assumes 200 as a baseline and does not add or subtract values correctly without an initial value of 200.

These values should not be confused with some safety feature. The Fitbit isn't going to magically stop you from going over any predefined value. In fact, using these auto generated values appears to make the HR sensor be off by a large margin. It seems that starting with a value of 200 allows the HR sensor to be much more accurate.

I think it would be acceptable to to input your max heart rate value as 220 bpm - your age to = your Max heart rate. You could set the lower limit to the same value as your low heart rate warning, as it would probably be acceptable too. You would just have to test these values.

[SunsetRunner]

As an update:

If you don't want to try the 200 40 values, you can check the boxes and use the values it initially displays which should be the max heart rate 220- your age which will = your max peak heart rate, It seems these autogenerated values are not used unless these boxes are checked.

hopefully this gives fitbit the info needed to fix this, then we can go back to leaving them unchecked.

[JohnnyRow]

I haven't done any experiments on this, but my understanding has always been that these values are used only for calculating heart rate zones, not for any actual data limiting.  Do be aware that changing these values definitely will change you heart rate zones, and with that, probably the limit of what intensity counts as Zone Minutes.

[SunsetRunner]

@SunsetRunner 

What I did was to set the custom max heart rate to my value. 220-age. Left the custom zone values alone (default values). By leaving the custom zone values alone, the heart rate zones determined by fitbit did not change. 

I did see the resting heart rate do fluctuate more often after changing the max heart rate value after syncing and rebooting the Sense. Lets see what happens when I exercise tomorrow morning if HR is more sensitive to heart rate changes.

Note that I didn't mention accuracy since I have no baseline or standard HR monitor to compare the Sense's reading to a known accurate HR. 

If this method actually works, makes one wonder why Fitbit does not make this information known to improve HR accuracy.

[deiG]

Sounds odd. Fitbit's software is really buggy.

[SunsetRunner]

@SunsetRunner 

@JohnnyRow 

an update...

changed the value of max heart rate and also changed the custom heart rate zones to 80% of max heart rate to get lower limit and 90% of max rate to get the high limit for this morning's exercise. Changing these values did not affect the default heart zones on my app. Also used a custom max heart rate of 166 (220 - age), changing this number did not affect the default heart rate zones as well. 

Wore the Sense 2 finger widths from my wrist bone as recommended in the help articles by fitbit for heart tracking accuracy. 

During exercise, I thought the heart rate tracked well during intervals. Heart rate went up and down in conjunction with my pace and breathing. I'm not sure if the Sense was able to pick up the spikes since the exercise was running intervals and not HIIT. Overall, what I found in the performance of the Sense whether using custom zones/custom max heart rate was the same with or without changing these values. 

As for accuracy, i'm afraid I can't say whether the Sense tracked heart rate accurately since I didn't have a chest strap (a good known standard) to compare what the Sense read as heart rate during exercise. 

Oh well, bottom line, I had fun during this morning's run....

[SunsetRunner]

I do want to post another update:

This morning I decided to remove my Fitbit Sense from my Fitbit account(After syncing the Sense), and I then did a factory reset on the Fitbit Sense. I then went though the set up process to re-add the Sense to my account, and then re linked it with my phone and updated it.

1. How to Remove Fitbit From Account - https://help.fitbit.com/articles/en_US/Help_article/1715.htm

2. How to Factory Reset Fitibit Sense - https://help.fitbit.com/articles/en_US/Help_article/2216.htm

After doing this, I noticed right off the bat that the SP02 Tracker app was not installed on my Sense despite it being installed out of the box. I also noticed that my resting heart rate while sitting at my desk is now in the 65-69 range instead of the 75-80 range. So this has definitively improved.

Another important point is I am using the default Spectrum Clock Face. I have a suspicion(can't confirm yet) that using the SP02 Clock Faces(perhaps others as well) cause the FitBit Sense hardware to lag. My Sense overall seems much more responsive with the Spectrum Face than it did with any of the Fitbit made SP02 Tracker Faces.

If my suspicions are correct, then any of the old Fitbit Versa 2/Ionic clock faces made by Fitbit that can be run on Sense should be Ok. I assume most of the testing Fitbit done with the Sense was done with the Spectrum Clock face, and as programmers know, with small devices like the Sense RAM, CPU resources are a premium, and the SP02 Faces just might need to be further optimized. I am wondering if perhaps the SP02 tracker and Clock Face together put too much strain on the Senses limited resources? Anyways just some thoughts.

Right now my resting heart rate seems much better than before. I'll walk on my lunch break and exercise at work and update everyone on my findings. I just wanted to share.

[SunsetRunner]

Update: Lunch Time Walk

The last few days, My brisk half mile walk on my lunch break that involves hills was only giving me a max of 109 HR and only 2 active zone minutes in a 9 minute walk.

Today after resetting and unlinking and re-linking my Sense as explained above and using the Spectrum Watch face, my afternoon walk today gave me 7 Zone minutes out of 9 total minutes, and a max Heart Rate of 131 up from 109 yesterday.

So it appears for my walks the issue is resolved and the HR monitor is where it should be or very close.

Will test spinning after work..

[SunsetRunner]

Did a manual count of my HR for one minute, Sense was within 1-2 BPM. I suppose that's somewhat positive info. Currently using a clockface with no SPO2 data. It doesn't strike me as practical to use a clockface with SPO2 values from your previous sleep since SPO2 only gets recorded while sleeping and will remain static during the day. 

Maybe next time I'll do a manual count of HR in between exercises just to see how the Sense is reading heart rates. 

Still need a known good HR monitor though to compare Sense readings from a known good (accurate) source to get something significant information on HR accuracy.

[SunsetRunner]

---

@SunsetRunner wrote:

I do want to post another update:

This morning I decided to remove my Fitbit Sense from my Fitbit account(After syncing the Sense), and I then did a factory reset on the Fitbit Sense. I then went though the set up process to re-add the Sense to my account, and then re linked it with my phone and updated it.

1. How to Remove Fitbit From Account - https://help.fitbit.com/articles/en_US/Help_article/1715.htm

2. How to Factory Reset Fitibit Sense - https://help.fitbit.com/articles/en_US/Help_article/2216.htm

After doing this, I noticed right off the bat that the SP02 Tracker app was not installed on my Sense despite it being installed out of the box. I also noticed that my resting heart rate while sitting at my desk is now in the 65-69 range instead of the 75-80 range. So this has definitively improved.

Another important point is I am using the default Spectrum Clock Face. I have a suspicion(can't confirm yet) that using the SP02 Clock Faces(perhaps others as well) cause the FitBit Sense hardware to lag. My Sense overall seems much more responsive with the Spectrum Face than it did with any of the Fitbit made SP02 Tracker Faces.

If my suspicions are correct, then any of the old Fitbit Versa 2/Ionic clock faces made by Fitbit that can be run on Sense should be Ok. I assume most of the testing Fitbit done with the Sense was done with the Spectrum Clock face, and as programmers know, with small devices like the Sense RAM, CPU resources are a premium, and the SP02 Faces just might need to be further optimized. I am wondering if perhaps the SP02 tracker and Clock Face together put too much strain on the Senses limited resources? Anyways just some thoughts.

Right now my resting heart rate seems much better than before. I'll walk on my lunch break and exercise at work and update everyone on my findings. I just wanted to share.

---

I just want to further update this post

I tested with Spinning today and the issue is resolved the Fitbit is now giving accurate heart rates during exercise. Following the instructions quote her should fix the issue for those having issues with theirs not tracking right during exercise.

Here is my vigorous Spinning Exercise BEFORE removing the Sense from my account and resetting it: (not accurate)

Now here is my Vigorous Spinning Exercise AFTER removing the Fitbit from my account, resetting it, and using the Spectrum default clock face: (accurate)

 As you can see there is a MASSIVE difference. 

If your having issues:

1.sync your Sense,

2.Remove it from your Account,

3.do a Factory Reset,

4.Install the SP02 Tracker app(not any of the SP02 clock faces) 

5. Use the default Spectrum Face or any of the Fitbit Faces that do NOT have SP02 on them.

You will then find everything on your Sense will work right. Heart rate will be accurate, sleep tracking will be accurate, resting heart rate, etc 

I know not having an SP02 face is not ideal, but those faces need to be further optimized, they cause the Sense and its HR to lag and not read properly, until its fixed, if you want your Sense to work right, this is what you have to do.

Hopefully this information will expedite a fix from the developers. 

[t.parker]

@SunsetRunner I'm curious, what is your reference you compare the results with to know the level of accuracy and whether you compare pictures (which in the app or on the Dashboard are not very accurate) or actual data? I never used SpO2 clockface and in fact, never used Sense for anything else but exercise tracking. 1 in 10 times it could be as accurate as PolarH10 chest strap (which is considered to be a gold standard of consumer HR tracking). Other times Sense never worked correctly (hence, I don't use it anymore). I've been comparing data (HR capture per second) against more than one sensor (PolarH10, PolarOH1). It was surprising when sometimes (when I actually expected it to fail) it was almost spot on (good enough for a wristwatch) and but I didn't do anything differently. Sense can track HR accurately but I see it being very sensitive to the noise (many times my running cadence was affecting readings, hence, I usually get higher HR readings on Sense than I should). How do you know your second Spinning tracking was accurate and the first wasn't? What are your data gathering and analysis methodology?

[SunsetRunner]

My references are:

1. Samsung Galaxy Watch

2. An Apple Watch Series 6

3.  Samsung Galaxy Watch Active 2

4. Built-in HR Monitor on my Spinning Cycle

Before the removing it from my account, resetting it, and changing the default clock face the Fitbit Sense would be 15-20 BPM lower than those 4 other sources. After removing it from my account, factory resetting it, and using the default Sperctrum Clock face, the Sense is now within 1-3 BPM of those 4 sources. Furthermore, the calories burned on the Sense is within 15 cals reported by my Spinning machine which was very close to what my other watches have always showed. 

Comparing the Fitbit Sense wrist HR monitor to a chest strap is simply ridiculous. The Chest strap is a medical grade device that has FDA approvals to be used in the medical industry. The Fitbit Sense is not a medical device, the HR Sensor in the Fitbit Sense is meant to give you an approximation of what zone your in. After this fix it does that acceptably. Before I done this, I could barely catch my breath and it was reporting 116 BPM, now i can barely catch my breath and its reporting between 145-150 (168 is max HR for my age) so an an aproximator(which is what the device is meant for, as are all wrist based trackers) its acceptable.

As you can see with Polar products they are FDA approved medical devices, comparing the Sense with such a device is simply unfair, and whoever came up with that idea is being dishonest. With this fix the Sense is just as accurate as Apple and Samsung which is acceptable for an on-wrist device.

https://www.accessdata.fda.gov/cdrh_docs/pdf3/K033646.pdf

[t.parker]

@SunsetRunner I disagree, the reference sensor should be as close to 100% of accuracy as possible, hence using PolarH10 because then I can tell whether the Sense or any other device (I test all the devices I get against H10 and OH1) is close enough to be acceptable for a) average use like estimating calorie burn, b) training with HR zones in mind, c) be used for HRV d) any other accuracy-related purpose. FDA is only in the USA so I couldn't care less about that (Fitbit sells devices also outside the USA). Wristwatches will never be 100% accurate as the wrist is pretty much the worst possible place to wear such a sensor but the accuracy can be measured with data and you can do all the analysis to find the best use of a particular sensor could be. When you do your analysis what is your data sample? For me, the snapshot of the roughly sketched chart isn't enough. It may be that I need very high accuracy on the go rather than average HR, and mind that average is rather meaningless without knowing standard deviation (two watches may come up with the same average HR but the output may be still very different - then are both of them accurate or none of them or one of them? Can't tell without std.dev.). What I mean is that if something is promoted as a fix and requires factory reset (it isn't unusual for Fitbit that factory reset may brick the device) then I'd rather see proper evidence based on actual data showing the improvement. I don't think developers will even consider that as any sort of useful input without backing up results with proof (although, Fitbit developers wouldn't do anything even if the proof would hit them in the face).

[SunsetRunner]

@t.parker 

great point...the HR readings on the Sense needs to be compared to a known good standard in order to get something meaningful in terms of accuracy. 

Then the question of why even use an HR sensor on a wrist wearable comes into question. From my experience, using the sense on Running/Walking exercise and observing the HR, it appears to be tracking HR changes well. Then again, I dont have anything to compare the Sense's HR reading. 

If only the Sense can pair to either an HR strap or HR armband via bluetooth/ANT+, that would be a great option.

[t.parker]

@SunsetRunner I use wrist HR for non-training activities. If I go for a hike or do short stretching/strengthening sessions I don't really care so much about HR to use a strap. For training, however, I use a strap because I need accuracy (very often I train with HR zones and I need my zones to be updated using lactate threshold and this requires accuracy). Currently, with the Garmin watch, I can use both - wrist and external sensors so that works for me. The wrist HR has its uses when you don't need great accuracy and in fact, most people don't need the accuracy of a chest strap. Fitbit users use HR mostly in connection with calories output and I can't imagine proper training with only three HR zones that are not really connected to any measurable training effect.

For me, Sense HR fails during runs, too. I have posted here plenty of comparisons showing how bad it can get. As I said, it seems to be very sensitive to noise and I "produce" lots of it when running with high cadence. I see it as a reason why, in my case, Sense registers higher HR than PolarH10 - because it mixes up the actual HR signal with my cadence (>180spm). The evidence of that was quite clear when my HR during runs suddenly spikes matching my cadence (I run with a metronome so my cadence is fixed at any given time). However, I recorded several occurrences of very accurate readings almost matching the chest strap. As a user, I didn't do anything differently, yet Sense magically provided very good HR. I can't explain it and it happens 1 in 10 times. I "retired" Sense about 3 weeks ago, got bored with testing it. I don't think it will ever be improved.

[deiG]

@t.parker , in general, is Garmin's optical HR sensor more accurate than Fitbit Sense? 

[t.parker]

@deiG in general - yes (I use Fenix 6 Pro), however, I wouldn't use it to train with HR zones. It's usually close but sometimes it just lags behind. Another reason for using an external sensor is that with PolarH10 the watch is able to monitor HR but also HRV during a workout. Accuracy of wrist HR for the casual users (non-athlete) should be enough but then there is a number of cheaper devices that provide similar accuracy. Sense can be this accurate, too but for some unknown reason, it's a real lottery. For sports tracking, is close but could be better so when I run the Garmin watch + PolarH10 + Stryd is my usual setup (and turning off wrist HR saves battery). When I buy a new watch, one of the key requirements (and this is really a must-have for me) that it must meet is being able to pair with external sensors so I don't need to be bothered whether the wrist sensor is accurate or not. Fitbit didn't figure out yet that such a feature could be a lifesaver for the company.

[SunsetRunner]

@t.parker 

Great posts!!! Thought I'd comment now in case your posts are removed by moderators for whatever reason the deem necessary in accordance with community guidelines. 

Also read a few disclaimers from other wearables about the challenges of HR accuracy taken from wrist sensors and the common denominator is for users that need HR accuracy is the recommendation of using chest straps. Won't mention the other manufacturers in case moderators remove this post due to perceived advertising. 

I'm considering armband HR monitors, it appears those provide accurate readings as well. 

Thanks for the great insights.