Testing Methods, Results, and Conclusions

This study verifies the accuracy of an Arduino-based heart-rate sensor as compared to a commercial PulseOx when subject to accelerating and decelerating heart rate patterns.  We attached a PulseOx and Arduino-based heart rate sensor to a participant’s finger and recorded their heart rate.  Heart rate was recorded before, during, and after the subject cycled on a stationary indoor cycling bike.  Refer to the identically titled testing protocol for full experimental details.  For the control, at rest, there was little difference between the two sensors’ measurement.  The average percent difference being 2.15%, averaged between all three subjects.  When subject to an accelerating and decelerating heart rate, the average percent difference was 7.46%% and 4.53%, respectively.  This study was conducted with three subjects.

Fig. 1. The stationary exercise bike which we used during the exercise. Fig 2.  The subject’s index fingers which were attached to the sensors were placed on the handle in the middle to reduce movement. Fig. 3-4. The subject cycles starting at the rate of 60 RPM. Fig. 5. The resistance was set to be level 1 throughout the whole experiment. The maximum speed reached was 120 RPM.

Fig. 6. The three graphs in the upper row plot the heart rate of the PulseOx sensor and SepSock sensor against one another while the sensors (attached to each hand’s index finger) are at rest.  This provides a control for the comparative accuracy of the two sensors throughout the experiment. Fig. 7. The three graphs in the lower row plot the heart rate of the PulseOx sensor and SepSock sensor against one another when the subject had an increasing and decelerating heart rate. Increasing heart rate was managed by cycling on a stationary indoor bike at a rate of 60 – 120 rpm. RPM were increased by approximately 10 rpm every ~60 seconds. A decreasing heart rate was then managed by allowing the subject rest for ~three minutes after exercising.

Fig. 8. Table shows the average percent difference between the heart rate measurements of the PulseOx and Arduino-based heart rate sensor under each condition for each subject.  The average of the subject’s percent differences are then shown.