Sensors and sophisticated algorithms together deliver check heart-rate variability and other exercise metrics.
Alisher Kholmatov, Recep Ozgun Micros, Security & Software Business Unit, Maxim Integrated
From weekend warriors to professional athletes, most people who engage in regular sports activity take note of various factors that influence their performance. Some meticulously track certain metrics from game to game to assess their progression (or lack thereof). Others are diligent about every aspect of their training and practice regimen. The emergence of sophisticated wearable devices like smartwatches and fitness trackers is bringing to light another important indicator of athletic performance: heart-rate variability (HRV).
HRV provides a measure of the time difference between successive heartbeats; the variation is controlled by the autonomic nervous system (ANS). This is the system that regulates our heartbeat, blood pressure, and breathing. Our ANS consists of two main components:
Our “fight-or-flight mechanism” (the sympathetic nervous system) is our response to external stress factors. This is what makes us energized and ready to face challenges.
Our relaxation response (the parasympathetic nervous system) is the opposite. This is what turns us toward a rest and recovery mode.
HRV tends to be low when we’re in a fight-or-flight mode and high when we’re more relaxed. High HRV is generally associated with greater cardiovascular fitness. The electrocardiogram test is the gold standard for assessing HRV. However, wearable devices with sophisticated algorithms are now making this data more accessible to more people.
Analyzing HRV for health insights
Because HRV provides insights into our ANS, it can be considered as the most important general wellness parameter. In the sports world, there’s a growing body of evidence pointing to the role of HRV in setting optimal training loads to enhance athletic performance. For example, a study of cyclists by university researchers in Spain and South Africa found that HRV-guided training led to better performance results.
HRV gives us some clues about potential cardiac conditions. Over time, higher HRV can indicate increased resilience, while lower HRV can point to chronic stress. HRV can be increased by changes in behavior: quitting smoking, losing weight if needed, exercising, and managing stress, for example. HRV analysis techniques that fall into three categories provide useful health insights:
Time-domain analysis, like standard deviation and root mean square (RMS), provides powerful differentiators of stressed versus relaxed conditions. Some examples include SDNN (the standard deviation of NN (or R-R) intervals), which provides a measure of changes in heart rate stemming from cycles longer than five minutes, and NN50, which notes the number of pairs of successive NN intervals that differ by more than 50 msec.
Frequency domain analysis shows the ratio of parasympathetic and sympathetic activity.
High frequency pertains to the parasympathetic system and the vagus nerve, which controls the parasympathetic nervous system
Low frequency pertains to sympathetic activity
Very low frequency pertains to the sympathetic nervous system, chemoreceptors, thermoreceptors, and the renin-angiotensin system (i.e., hormones)
Non-linear analysis can point to underlying cardiac conditions and includes:
Detrended fluctuation analysis (DFA), which looks for self-similar patterns by analyzing the power spectral density (PSD). Peaks in PSD indicate repetitive patterns.
Entropy analysis, a measure of randomness over time. Decreased HRV and increased randomness of HR are independently associated with high-risk conditions.
Poincaré plot analysis utilizes scatter plots of consecutive pulse interval points. Consecutive pulses that vary by large amounts will have larger scattering around the diagonal. Low HRV will shrink and cluster around the diagonal. Unbalanced HR behaviors such as fast acceleration and slow deceleration will generate asymmetric plots. Large off-diagonals show skipped heartbeats, which usually indicate an arrhythmia problem.
Understanding exercise metrics
Let’s now take a look at exercise physiology and associated metrics. According to the American Heart Association, cardiorespiratory fitness is a better indicator of mortality than any other risk factors, including smoking, hypertension, and high cholesterol. The current gold standard to determine cardiorespiratory fitness is the VO2 max test, which tests for the maximum amount of oxygen that one can use during intense exercise. VO2 max is related to peak endurance. The test is generally administered in a medical facility using a treadmill and an oxygen mask. Other important exercise metrics include measurements of HR and recovery and of excess post-exercise oxygen consumption (EPOC).
Based on the energy expended during exercise, HR goes up to deliver the oxygen needed for energy production. Max HR is the upper limit. The max HR scale is divided into five zones, based on exercise strength or the exercise goal. The target heart rates are age-dependent. Zone 1 is the very low intensity zone (50-60% of max HR). Training at this intensity will boost recovery and get you ready for higher zones. Zone 2 is essential for a runner’s program (60-70% of max HR). Exercising at this zone will improve general endurance. The last three zones, however, are the most interesting:
Zone 3 is the aerobic level (70-80% of max HR). This is a comfort zone, where oxygen is readily available, and energy is generated mostly from burning fat.
Zone 4 is the anaerobic level (80-90% of max HR). This is out of the comfort zone, where the goal is to hit the maximum oxygen consumption (VO2 max) and remain there as long as possible to continuing burning fat post-exercise.
Zone 5 (more than 90% of max HR) is not recommended, as this is not at a healthy level and can trigger long-term health effects.
Recovery time after exercise depends on the total O2 deficit. EPOC refers to the amount of extra oxygen needed to recover after exercise. This extra oxygen consumption actually burns additional calories after exercise. The recovery itself has a fast and a slow component. During the fast component, the muscles return back to their normal state. During the slow component, the lactic acid that is generated during exercise (as glucose is burned to generate energy) is removed from the muscles. EPOC lasts up to 48 hours depending on duration and intensity of the exercise.
Besides the gold-standard VO2 max test, there are other well-defined protocols that can guide exercise. Some are geared toward establishing optimal training processes, while others are for improving the oxygen consumption rate and still others are aimed at increasing post-exercise energy consumption. During the exercise, VO2 increases as the workload increases. However, at some point, despite an increase in workload, an individual will reach his or her VO2 max. VO2 can be improved with training. Cross-country skiers, runners, and swimmers tend to have the largest VO2 max.
According to Market Reports Hub, the smart sports and fitness wearables market is projected to reach $14.9 billion worldwide by 2021. These are the very devices, the analysts note, that are about delivering meaningful data that can turn into actionable information that helps people improve athletic performance or manage overall personal fitness. Indeed, there are now several wrist-worn and even in-ear devices on the market that measure HRV as well as the exercise metrics previously discussed.
While their HR measurement accuracy may not be as precise as measurements collected via an ECG based chest-strap device, they provide an indication of VO2 max as well as of parameters like HRV and EPOC. A variety of HRV-based guided exercise applications is also now available. All of these tools provide guidance to help users optimize daily stress load and recovery, personalize training plans based on the individual’s response to stress, plan for periods of rest between activity, and make adjustments to enhance athletic performance.
Wearables rely on these key components to deliver accuracy in measuring HRV: proper optical and industrial design, high signal-to-noise ratio (SNR) optical bio-sensors, and, of course, advanced algorithms. From an optical design perspective, because measurements are based on the interaction of light with skin, it is important to consider factors such as crosstalk suppression, separation distance between the device’s LEDs and the photodiode, and opto-mechanical integration.
As HRV is mainly dependent on small beat-to-beat changes in HR periodicity, the most important input here comes from a reliable HR reporting device. The signal quality of this device, whether based on electrocardiogram (ECG), photophlethysmography (PPG), or an acoustic approach, is a limiting factor on accuracy. Bad HR readings can derail HRV measurements and subsequent VO2 max, recovery, and EPOC estimations. The robustness and high signal-to-noise ratio of ECG and PPG sensors used in the reporting device can alleviate the impacts of optical noise on accuracy. Industrial design must factor in where the HR readings will be captured—even elasticity of the wrist strap is crucial.
Let’s now spend more time discussing the algorithms. To support HRV-guided training, the algorithms in these wearables must overcome five foundational challenges that impact accuracy of optical HR measurements:
Optical noise. Algorithms with capabilities like ambient light rejection and “picket fence” detect-and-replace can, respectively, reduce the undesirable noise and changes in ambient light conditions that hamper accuracy.
Impact of sensor location on the body. Muscle, tendon, bone, and overall arm and wrist movement can generate optical noise that impacts measurements. Another factor is the signal response at different wavelengths. Designing a system that consists of multiple sensors and light sources is key to overcoming these challenges. Also, while HR can be tracked during exercise, the best VO2 and recovery estimates can be made at resting states before and after exercise. From an algorithm standpoint, effective algorithms are designed with the ability to utilize the specificity of the sensor location on the body and to compensate for any related optical noise as necessary.
The effects of skin tone. Biological factors such as the darkness of skin, the presence of body hair, and even the presence of tattoos can make it more challenging to capture HR measurements optically because they impact light absorption and, thus, signal quality. Algorithms that account for poor signal quality are needed.
The effects of low perfusion, which is an indication of pulse strength and can also be triggered during measurement by low body temperature at the sensor location. Algorithms that account for poor signal quality are needed to cover these cases.
Motion compensation, which includes the “crossover” problem, where pulse rate and motion frequency cross over each other when the subject is in motion, negating one of the measurements. Algorithms must be smart enough to differentiate between motion and HR modulation.
HRV is an important indicator of health, well-being, and general fitness. In addition to aiding in sports coaching and athletic training, HRV monitoring also has applications in areas including stress maintenance and sleep analysis. Health-monitoring wearables equipped with sensors and sophisticated algorithms that together deliver HRV and other exercise metric insights are enabling us to understand, in real time, how we can adjust our training regimens, optimize performance, and manage our fitness goals.