How accurate are heart rate monitors on fitness bands?

There has been a recent trend among fitness trackers to include a heart rate monitor (recode). Wristband wearables from Fitbit, Jawbone, and eventually Apple (i.e. the future Apple Watch) have this capability. Measuring your heart rate at rest or while exercising can provide valuable information about cardiovascular fitness.

The basic principle of these monitors is that LED light is directed toward a patch of skin, where it can penetrate the surface and get reflected or scattered by the blood underneath. Sensors next to the LEDs can detect the light signal from the blood (e.g. red glow when you shine a flashlight on your finger). Your pulse generates an increase or decrease in the volume of blood in vessels giving rise to a periodic signal that can be quantitated. This principle is used by the Xbox and Samsung Galaxy S5 to measure heart rate.

One big question is how accurate are these wearable heart rate monitors? Sharon Profis at CNET did a nice study comparing some wristband heart monitors to the gold standard, an EKG machine.

Five devices were tested: (1) Garmin VivoFit, (2) Basis Carbon Steel, (3) Withings Pulse O2, (4) Samsung Gear Fit, and the (5) Samsung Galaxy S5 (smartphone). The Garmin device comes with a special heart rate monitor that is strapped to your chest and measures the electrical signals from your heart just like an EKG. Devices (2)-(4) are the wristband fitness trackers, and the Galaxy S5 is a smartphone (you place your finger over the flash to measure heart rate).

Ms Profis measured her heart rate both at rest (80-90 beats per minute) and after running (160-170 BPM) on a treadmill for all the devices as well as the EKG machine. The measurements were done three times, averaged, and then compared to the EKG numbers. The percent deviation from the EKG standard was recorded (see Table 1).

All the devices performed admirably for the at rest heart rate with an error between 3% to 10%, which is acceptable. On the other hand after running, the bracelet monitors scored much worse with errors over 50% and in the case of the Gear Fit, not providing a reading. Interestingly the Samsung Galaxy S5 performed quite well. Not surprisingly, the Garmin device with the special chest strap showed low error too.

One possible explanation for the inaccuracy at high BPMs is that the device may be shifting or jostling around when you are running. However as Ms Profis mentioned, the devices specify that the user has to be still to get a reading ("no talking, no moving, no muscle-tensing, no sweating, no smudging allowed"), and so the measurements were performed after she had stopped running.

The most likely explanation is that at faster heart rates it becomes harder to distinguish signal (actual beat) from noise (random fluctuations in signal); a higher frequency, lower amplitude signal (i.e. higher BPMs) begins to resemble noise. One possible reason why the Galaxy S5 did well at the higher rate is that the signal is stronger (i.e. more blood closer to the skin) in the finger tip than in the wrist.

In conclusion, the wrist heart monitor devices provide reasonably accurate measurements at lower heart rates, but not at faster rates. The accuracy of the Apple Watch remains to be seen. Surprisingly, the finger sensor on the Galaxy S5 performed almost as well as the dedicated chest-strap heart monitor of the Garmin device (which can be worn and used when you are running). Of course, one should not forget that there is an accurate low-tech option which is to measure your pulse manually (link).

Table 1. Error from 5 heart rate monitoring devices compared to EKG standard at two different heart rates (CNET).