Hi Have been in persistent AF before cardioversion in August22 and enjoyed a lovely 9mths of NSR till it kicked in again (back on the waiting list..)
In checking my heart rate regularly using a Kardia Mobile I became suspicious of the 'calculated' BPM declared on the app and started manually checking them. I have attached a picture of one such scan, declared as 117bpm but showing only 50 QRS peaks over the 30sec sample period i.e. average of 100bpm! I can accept a discrepancy of a couple of bpm but 17!?
I assume that the tick marks below the ECG trace are peaks identified by Kardia (?) used in the calculation algorithm which correspond one to one with the QRS peaks I used for my count, so why should the result be different?
I've forwarded to AliveCor for comment but no response!
Any thoughts/similar experiences?
Written by
ObiDyer
To view profiles and participate in discussions please or .
Could the Kardia be picking up ectopics? This was given to me as an explanation for discrepancy between my smart watch and manual pulse readings. My GP recommended Kardia 6l and when I took a printout he commented it was better than the printouts from the practice ECG. Even though my watch showed elevated readings the alerts were accepted as evidence of issues which subsequently merited an ECG implant (LINQ) which did it's job and picked up AF/AFl which has not showed on any previous 12lECG or Holters.
Picking up ectopics is a requirement for a heart rate monitor to be of any use for detecting arrhythmia's. Its why taking your pulse by hand via the wrist always gives a lower than reality heart rate. The Kardia does detect ectopics but relies on an algorithm taken from the last 5 secs of reading only which for AF is in all probability going to be wrong!
So I got 98 bmp, but here in the US it's early so haven't had my coffee yet My Kardia is usually very accurate when calculating heart rate, even in afib, but sometimes not when in afb and when inaccurate I too have noticed an overcount -- so always best to count the QRS intervals like you did. That's what ep's do, even with their 12 leads.
The important thing is that Kardia is generating a very clear ekg, showing you in afib, that you can share with your cardiologist or GP. It's incredible that such a small and reasonably priced device can produce their very accurate medical grade ekg's! I also have the 6L and keep a Kardia Card in my wallet as a back up.
The automatic determinations and HR calculations are secondary and should not be absolutely relied on.
Just for info, ViATOM Pulsebit-EX ECG Monitor is the replacement I now use after discovering the Kardia AF inaccuracies some weeks ago. It actually produces much clearer strips, accurate numbers, has a PC app with measurement tools and prints to pdf files without any subscription. Also it has its own screen display.
Thanks. Not familar but will look into them. Do either have REAL TIME afib alarms?The advantage with Kardia is that at least here in the US, ep's accept the ekg's for decision making. Absolutely nothing wrong with quality or accuracy of Kardia's ekg's. The auto determinations are another story and frankly ep's don't even look at that. They just look at ekg and do their own analysis.
The Viatom Pulsebit flags AF as a red banner 'Irregular Heartbeat' and if you download it to the pc app gives a high quality graph along with HR (correct number) QRS and QT/QTc. Her in the uk my cardiologist is not too impressed with any 'home' ecg and will only look at the 12 lead hospital results.
My own concern with the Kardia is that knowing that the 5 sec algorithm is probably reporting inappropriate heart rates for AF, the 6 lead strips are also produced by algorithm from 3 contact points and so concerned me that it perhaps had the potential to be similarly inaccurate on the extra lead interpretations, which was really the only reason I had for buying the 6l version.
In the US, most ep's accept and make treatment decisions based on Kardia. Not necessarily true with regular cardiologists, however. who have less ekg training and tend to rely on auto algorithms more.
Actually all of Kardia's determinations are only made from lead 1, so the 6L,which I have, has no advantage here. The only advantage is that a six lead ekg can sometimes be more helpful than a single lead in distinguishing certain arrythmia's such as aflutter, atach, from one another. But again, most ep's do not bother looking at alogorithm's, they just want to see the ekg and Kardia's ekg's are very clear and the ep's I've spoken to prefer them over the ones they get even from the Holter monitors.
I'm still looking for a wearable monitor with real time afib alerts, but still haven't found one. It may be a regulatory issue because the technology is there.
I'm also of the opinion that my Kardia is not showing the correct Bpm. I've been using the device for very many years. This is in fact my second device. I became very aware of this when I went into fast AF 10 days ago and was in A&E. The Kardia was showing a much lower bpm (average) than the correct one, discovered when I presented at A&E. I wondered why I felt so rough on what I thought was not a terribly high rate.
I haven't been in contact with Kardia yet but will do so when I can summon the energy !
Interesting, my Kardia seems to over estimate the bpm. I'm trying to manage my meds (under medical supervision) to keep my heart rate sensible/safe while avoiding the horrible side effects if I crank up the sotalol too much until I can be cardioverted so knowing my bpm reasonably well is significant for me. Sounds like manual counting beats the 'tech'!
I'm waiting to replace my battery although it shouldn't really need it. In addition there is a buzzing noise coming from the actual finger plates so that it is unreadable on many occasions ....massive interference. I really do need to speak to Kardia !!
Definitely let us know what they say, but don't go through the bother of an exchage if they offer, because they are all like that Certainly accurate enough to tell if you're in fast afib or not, and easy enough to count the beats yourself from the ekg. It's really the ekg that shines with Kardia and that's the most important thing. A good EP never even looks at what the machine tells him about bpm and diagnosis from their 12 lead -- they look at the ekg and will do a manual count if they think it's important.
I calculated the rate at between 150 to 60 bpm. That is using the 300 / number of blocks between 2 consecutive R peaks. My average over the 30 seconds is 105 bpm by picking lowest and highest only.
We don't know what algorithm Kardia uses to do the calculation. Your's or mine or some combination.
their calculations are accurate. It showed Afib and the rate was not too fast. There is no accuracy in computing the rate while in Afib, just approximately over a minute as average since the rhythm in Afib by default is irregular.
For AF It would equal the number of irregular beats over a one minute reading.
If you choose a 30 sec scan then multiplying x2 may not be completely accurate but usually close enough. The Kardia calculates based on the last 5 secs, which for sinus rhythms or occasional pacs/pvcs etc may be a close approximation but for AF with large rate variations is likely to be completely inaccurate.
The Kardia user manual shown on the British Heart Foundation page bhf.org.uk/-/media/F3570E1C... quotes the following.
Heart Rate
"During your EKG recording, your real-time heart rate will be shown. When reviewing previous EKGs, the average heart rate taken during that recording is displayed. Heart rate is calculated as the time interval between consecutive heart beats; or more specifically as the inverse of the time interval between consecutive R-waves in your QRS complex. During an EKG recording, the current heart rate is measured from an average of this inverse calculation over the last 5 seconds. For stored EKGs, the average heart rate is the average of this inverse calculation over the entire 30 seconds of the recording"
I only discovered this when I bought the Kardia 6l and recorded complete inconsistencies in the heart rate figures displayed, some times more than 30% wrong as it is totally dependent on counting the last 5 secs. One of my first strips read gave a 138 result but if you count the pulses (conveniently marked for you on the strip) they only added up to 108. I also discovered that if you select a 60 sec scan which should give a more accurate picture of you HR it actually only gives you the first 30 secs of the strip doubled to give an even more inaccurate HR result.
I have a number of 60 sec strips that are completely wrong when compared to counting the actual beats. Always too high, example Kardia result HR 124 actual count 92.
I also discovered that if you select a 60 sec scan which should give a more accurate picture of you HR it actually only gives you the first 30 secs of the strip doubled to give an even more inaccurate HR result.
Based on this the Kardia would seem to be completely unsuitable for keeping track of AF particularly if you are susceptible to large range of rates that need to be kept under control
Update... rebooted my phone (old Moto G7 used solely for Kardia which won't run on my newer G9!!) and the algorithm rate seems to be in much better agreement with manual counting! Will keep cross checking with manual counting but maybe it's just a bit of drift in the algorithm constants that needed resetting etc?
Me again Obi: I think you'll find that regardless of the setting of the algorithm, if you are in NSR with normal rhythm, the HR measurement will be spot on, or very close to it.
The HR irregularity component of an afib episode makes an "accurate" measurement of said HR impossible as it depends on the particular time frame used for "the count". A 5 second window would potentially give different HR from a 15 second window (manual wrist pulse), which could give a different HR measurement from the 30 second manual beat counting from your Kardia trace. The longer the sample time of an irregular event, the more accurate will be the count of events. Counting your pulse via wrist or carotid for an hour would give a very accurate measurement of "average" HR.
BUT.....what difference does it make? You can look at this trace and see you are in afib-the beats are irregular, the HR is elevated, lack of P waves is confirmation. If you REALLY need a more accurate HR measurement, take 20 Kardias in a row and average them out! (I know I'm being silly, but I think you see my point that reliance on just a small segment of data can oft times be misleading, but not necessarily dangerous.)
My experience of a number of years measuring my PAF rates to ensure they don't go too high during strenuous activity has resulted in generally accurate 'average' pulse counting during 1 min of reading. I have used 3 different home ecg units, Viaton, Wellue and Vion until recently trying the Kardia 6l. The first 3 give an accurate count ie 100 pulses on the strip printout (including pvc's if present) will give a reading of 100BPM. If I go for a Cardiology check up, they take a 30sec reading and the HR rate they report is the real time result. There is no attempt to work out an impossible average over a random time period. If I can feel what may be a high AF rate during some kind of exertion or stress etc then I need to know what that real time number is so that if excessively high I can rest until the actual number of pulses has fallen to an acceptable level. That's why its important to have an accurate correctly counted number displayed. Pulse counting is a simple basic task. As a retired electronics design engineer, former head of advanced design at a division of Marconi I can assure anyone that a piece of electronics that is unable to fulfill this requirement should be regarded as unacceptable and not relied on for reassurance!
I read somewhere in a post on this site that you can modify the time period of a Kardia reading, taking it up to several minutes. Don’t shoot me if I’m wrong! It may only be on newer models.
I have a Wellue AI ECG wearable monitor that counts all beats and gives averages, mins and maxs by the hour. Reports are pretty comprehensive but of course it’s not accepted by the medical profession (yet?). You do have to download the data so it’s not immediate reporting but as the AI bit suggests they review data on their server to report back on a number of potential arrhythmias. It’s only slightly more expensive than a Kardia 6L.
You are correct you can set the Kardia for longer readings. The results for me were that the longer the time period the less accurate the result presumably because its still only counting based on the last 5 secs. No problem for sinus rhythm but not going to give a correct answer for AF. I also use a Wellue Pulsebit EX which gives completely accurate pulse counting.
"The algorithm in the KardiaMobile application calculates heart rate by not only counting heart beats, but also by looking at things including morphology to give you a best “calculated average”, it will not always be very close to the preliminary number that is shown while recording the EKG. This number may also be different from the number shown by a different product from a different company, that is also measuring beats per minute. This is because other products are only measuring the beats per minute, and are most likely not taking other factors into consideration like the morphology and other EKG data."
All sounds a bit fluffy/waffly (aka marketing speak) so I've asked for more specific details eg what is 'morphology', what other data is included, if it's not 'beats per minute' what is it (and why is it quoted in BPM!) etc etc etc...
Response to my request for more detail: "Unfortunately, we are not medically trained and are not able to answer medical-related questions, as we are only technically trained. We advised seeing or contacting a medical specialist for further explanations."
As I didn't ask for medical info, only technical stuff around data processing, I can only assume they don't know or have something to hide!
Seems they have access to someone else who suffers equally from waffly marketing speak...
Thank you for your patience. We've got a response from our business office team. Below are the answer to your questions.
What is meant by 'morphology'? In the context of an EKG, 'morphology' refers to the shape and form of the EKG waveform. This includes the P wave, QRS complex, T wave, and other components of the cardiac cycle. Changes in EKG morphology can often indicate various heart conditions.
What 'other EKG' data is included? We can't provide any additional data about our algorithms and what is included when calculating the BPM.
Why are these other data important? Same answer as above, we can't provide additional information on what factors are included in our algorithm.
Why is the Kardia output stated as being BPM (beats per minute) if it isn't? This is still a BPM.
Why does the Kardia trace include 'tick marks' to highlight significant factors, but bear no relation to the calculated figure? We are not sure what tick mars the customer is referring to. We have tick marks that assist in calculating the BPM my marking the R wave in a QRS complex. We also have tick marks that are included in our Advanced Determinations as a part of our KardiaCare subscription. These additional tick marks indicate when the Kardia algorithm has detected PVCs and PACs
As mentioned already by some, the irregular heart rate in AF means that the calculated rate will vary depending on which part of the tracing the measurement is taken from. I'm not sure what your practical concern is. You measured over 30 sec and got 117, but if you measured your heart rate over 45sec, or 1min, or 2min (or any diffent time period), you would get a different number each time. That's just a consequence of the irregular nature of AF. A ball-park heart rate figure is good enough for most purposes.
For the SAME SET OF DATA why does Kardia's algorithm state the heart rate to be 117 when a manual count indicates that to be in error by approaching 20%!?
Nothing to do with sample times just simple arithmetic..
If my GP wants my heart rate below 100 and I tell him it's 117 (when it's really only 100) I am taking excessive medication for no reason and degrading my quality of life with it's side effects!
It has everything to do with sample timing and duration.. Count your pulse for a minute whilst in AF. Then count it again for another minute. You will most probably have two different numbers. Which is your heart rate?
In AF the heart rate is continually varying.
Counting for a longer time, such as you have done over 30secs, will give a better estimate of overall heart rate than the kardia number (which apparently relies on a count over 5 seconds). I agree that it would be useful if the kardia took the count over the full tracing time .
ObiDyer, I have the exact same problem with the Kardia 6L. If your heart rate is regular, the heart rate shown in the summary report is correct. If your heart rate is irregular, the heart rate is incorrect. I have an irregular heart rhythm and the Kardia 6L always over-reports my heart rate and is not dependable.
As an example, I recorded a 1 minute ECG. Kardia displayed a heart rate of 128 BPM on the resulting summary report. When I count the number of R waves on the 60s recording, I get 96. Apparently, Kardia is only using a small subset of the ECG to calculate heart rate, which would lead to inaccuracy if the heart rate is irregular. When I use the EMAY 6L monitor, the summary heart rate matches with the number of R waves counted during the ECG recording.
To me, Kardia's method for calculating heart rate is flawed. I can accept an min and max heart rate but over-reporting heart rate based on a truncated time interval is unacceptable. I will contact AliveCor and see whether I can get a response. My expectations are low based on the response that was given to your inquiry.
Content on HealthUnlocked does not replace the relationship between you and doctors or other healthcare professionals nor the advice you receive from them.
Never delay seeking advice or dialling emergency services because of something that you have read on HealthUnlocked.
My Kardia is usually very accurate when calculating heart rate, even in afib, but sometimes not when in afb and when inaccurate I too have noticed an overcount -- so always best to count the QRS intervals like you did. That's what ep's do, even with their 12 leads. 
Accuracy of Kardia
Kardia showed normal sinus rhythm with wide QRS
Kardia readings Tachycardia, followed by AF
Kardia 6l and old Kardia together
strange ECG on Kardia
