Using Heart Rate Variability for Health Monitoring

Can measuring heart rate variability with your smartphone predict the onset of an illness?

As the COVID-19 outbreak continues to spread, the FDA has warned consumers against buying at-home testing kits for SARS-CoV-2. As of March 23, the Food and Drug Administration hasn’t authorized any self-administered tests. If you suspect you have COVID-19, contact your healthcare provider and follow their instructions.

Still, if that doesn’t alleviate your concern, you can try an additional science-based method to track your physical well-being — heart rate variability. A health marker that you can measure with your smartphone, heart rate variability can help you spot the onset of an illness before you feel any symptoms and monitor your recovery if you get sick.

What is heart rate variability?

Heart rate variability (HRV) is a measure of variations in the time intervals between your heartbeats — how “uneven” your heartbeat is. First designed to track astronaut performance in the 1960s, this metric has now become well-studied. Scientists use HRV for different purposes, for instance:

• to measure stress levels
• to separate healthy people from those with chronic health conditions
• to assess the immune system
• to predict the speed of recovery after a severe illness

All of this makes HRV a valuable tool for tracking your physical well-being. For example, you can spot the onset of illness — including COVID-19 — before you develop symptoms.

HRV can also help you understand how well your body is coping with being sick. If your scores get progressively worse, it can be a signal that you should call your doctor or even go to the ER.

Finally, HRV is a good way to monitor recovery. Healthcare professionals could potentially use HRV readings to identify which patients need immediate intervention, evaluate how effective treatments are, etc.

How does HRV work?

As we’ve already mentioned, HRV measures how uneven your heartbeat is. This variability is a good thing — your heartbeat is not supposed to have an even rhythm. Higher variability shows that your nervous system is good at adapting to changes inside and outside of your body. When you start breathing faster, take a sip of coffee, or have a stressful thought, your autonomic nervous system sends signals that adjust your heart rate. And since everything changes all the time, your heart never beats evenly.

The higher your heart rate variability, the better your body is at adapting to changes and stressors. When you’re overwhelmed by stress or are in bad shape due to lack of sleep, alcohol abuse, overtraining, or respiratory illness, your HRV begins to drop. This means your body is having trouble adapting to changes and maintaining homeostasis — a stable internal environment. It’s a signal you need to take measures and regain balance.

How can you measure HRV?

There are two ways — you can use the camera on your smartphone or a special heart rate monitor.

The Welltory app uses the heartbeat data tracked through the camera or heart rate monitor to calculate different HRV metrics, using time-domain analysis and frequency-domain analysis.

Key time-domain metrics include RR, SDNN, pNN50, and rMSSD. All of these metrics essentially estimate how heart rate variability changes over time to evaluate the balance between your nervous system’s sympathetic and parasympathetic activity.

The two most widely used measures of HRV are SDNN and rMSSD. They’re calculated based on RR intervals — the time intervals between heartbeats. SDNN is basically the standard deviation of a measurement’s RR intervals, while rMSSD is the root mean square of successive differences. Research shows that SDNN provides a more accurate assessment of sympathetic activity: the activation of “fight-or-flight” in response to stress, the ability to adapt to changes and continue functioning and being productive. rMSSD is a measure of parasympathetic activity — how well you are able to “rest-and-digest” and recover from stress.

There is an additional type of analysis that might help you understand which process is now dominating in your body — stress or recovery. It’s called frequency-domain analysis: it uses metrics that are calculated by transforming the length of each RR interval into waves to measure their frequency. The resulting waves are then divided into three bands: low-frequency (LF), high frequency (HF) and very low frequency (VLF).

Welltory calculates the share of each power spectrum and the sum of all spectra, or total power. But we use these metrics with caution. Research has shown that frequency-domain analysis can be highly inaccurate: for instance, taking a deep breath affects LF power and consciously controlling your breath affects HF. Plus, these metrics often vary significantly throughout the day.

This means that you need to stick to very strict guidelines if you want to draw meaningful conclusions from frequency-domain analysis:

• take measurement lying down
• measure at the same time every day
• lie down for at least 5 minutes before measuring
• use the bathroom before measuring
• don’t measure if you’re on your period

If you stick to all of these guidelines and notice significant changes in your frequency-domain metrics, then we might — cautiously — draw some conclusions about the state of your autonomic nervous system.

How can you use HRV to spot the onset of illness?

We’ve already established that HRV metrics might fluctuate when you’re sick. But how exactly can you spot the onset of COVID-19?

Scientists are currently trying to find biomarkers that could help identify people with SARS-CoV-2 even before they develop symptoms. One biomarker they’ve been able to pinpoint is the C-reactive protein (CRP) — a marker of systemic inflammation in the body. Clinical data from infected patients has shown that CRP in COVID-19 patients changes before other blood test parameters — a sign this marker could be useful in detecting the first signs of infection:

“The clinical data of 33 patients with respiratory symptoms caused by the novel coronavirus in Wenzhou city from January 15 to February 12,2020, were thoroughly reviewed. According to the patients’ histories, the patients were divided into two groups: those who spent time in the main epidemic area and those who did not spend time in the main epidemic area. <…> Respiratory tract ailments and systemic symptoms were the primary symptoms of novel coronavirus infection in the secondary epidemic area; these symptoms are not typical. The abnormal increase in serum amyloid protein (SAA) may be used as an auxiliary index for diagnosis and treatment. CRP changes before other blood parameters and thus may be an effective evaluation index for patients with COVID-19 infection,” — Analysis of COVID-19 clinical features in secondary epidemic area, Weiping Ji, Gautam Bishnu, Zhenzhai Cai, Xian Shen, published on March 13, 2020.

Problem is, you can’t do blood tests every day in order to track changes in your CRP levels. That’s where HRV measurements come into play — there are a number of studies that prove there is a negative correlation between CRP levels and HRV metrics:

“Hamaad et al. (2005) tested the association between time and frequency domain indices of heart rate variability and circulating IL-6, high sensitivity CRP (hs-CRP) and white cell counts, in a sample of 100 patients with proven acute coronary syndrome. In addition, they compared these metrics with healthy controls (n = 49) and estimated possible relationships on repeated measures at 4 months in recovery (n = 51). They found modest negative correlations between all inflammatory biomarkers and mainly SDNN, VLF and LF power. The strongest associations were seen between WBC and SDNN (r = −0.351). However, relationships did not persist on multivariate analyses after a 4-month period. According to the authors, the correlations were observed largely among heart rate variability indices reflecting sympathetic activity, suggesting that the inflammatory response in acute coronary events may be associated with sympathetic activation instead of vagal withdrawal.”

“In another study, Lanza and colleagues (2006) assessed heart rate variability and measured CRP serum levels within 24 h of admission in 531 patients with unstable angina pectoris. They found a significant negative correlation between CRP levels and all heart rate variability metrics derived from both time and frequency domain, with the highest correlation coefficient with SDNN and VLF. After categorizing patients into 4 subgroups according to CRP quartile levels, significantly lower heart rate variability values were found in the upper CRP quartile. The subsequent multivariate analysis revealed that SDNN and VLF were the most significant predictors of increasing CRP, whereas CRP was a strong predictor of impaired ANS activity as well.”

“In a similar study (Nolan et al., 2007), a negative correlation between CRP and heart rate variability frequency components was reported, whereas a decreased HF power (reflecting vagal tone) in the high CRP quartile, compared to the lowest one, was found.”

“Psychari et al. (2007) also reported a strong inverse association between CRP and several heart rate variability indices (SDNN, HF, and LF).”

“Recently, von Känel et al. (2011) investigated the association between heart rate variability measured in the time domain, CRP, IL-6 and fibrinogen, in a cohort of 862 subjects recruited from the Heart and Soul Study, which assessed health outcomes in 1.024 outpatients with stable CAD. They found that SDNN was inversely and significantly associated with inflammatory indices, after adjustment of all covariates.”

To sum up these findings, SDNN has the strongest correlation with increased CRP levels. If you notice your SDNN falling for several days in a row, this may be a sign you’re getting sick. Same goes for rMSSD.

Another worrying signal is a decrease in HF power: it reflects your vagal tone and tends to fall when CRP levels are rising. Lowered LF and VLF power values can also be indicative of a spike in CRP levels.

In other words, altered HRV measures during the COVID-19 incubation period should be reflecting the stress your body is experiencing because of the developing inflammation.

Keep in mind that heart rate variability readings can fluctuate throughout the day. To eliminate these fluctuations and ensure you’re monitoring trends from one day to the next, take measurements at the same time every morning and every evening in the same position. HF, LF, and VLF values are even more sensitive than SDNN and rMSSD, so they must be interpreted with extra caution.

TL;DR: How can HRV help you spot the onset of COVID-19 before you develop symptoms?

Look out for the following signs:

• SDNN and rMSSD values drop for several days in a row
• For 300-beat measurements taken while lying down, HF, LF and VLF values drop for several days in a row
• pNN50, on the other hand, may either stay intact or also start dropping
• If you spot these changes in your HRV measurements, you can also get your CRP levels tested — elevated levels could be an early sign of an infection, which may or may not be COVID-19.
• Seeing a drop in HRV metrics and elevated CRP? It’s best to self-quarantine for at least two weeks. If you need to leave the house, wear a mask to prevent the asymptomatic spread of the virus — you might be asymptomatic or still be in the incubation period.

What if you already have symptoms?

Let’s recap. The three most common symptoms of COVID-19 include:

• fever
• dry cough
• shortness of breath

Other potential symptoms include:

• sore throat
• headaches
• fatigue
• body aches
• diarrhea (rare)
• stuffy nose (rare)

Runny nose and sneezing are NOT usually seen in COVID-19 patients. If you develop these symptoms, it’s likely a common cold or an allergy.

Suppose you’ve developed symptoms associated with the novel coronavirus. How can HRV help you?

You’ve been tested and you are positive for COVID-19. Your doctor told you to stay at home and monitor your symptoms.

Measuring your HRV can help you track the course of your illness and notice if things get worse.

The most common complication of COVID-19 is pneumonia. It can further lead to acute respiratory distress syndrome (ARDS), lung failure and even death. So it’s important to closely monitor your condition in order to spot the first signs of pneumonia. HRV can help.

Researchers have compared patients with pneumonia with healthy control subjects. They’ve noticed that having pneumonia decreases your total power and VLF power by 3 to 5 times. If you have pneumonia, your LF power also drops compared to your normal range. Most importantly, people with pneumonia have an LF/HF ratio lower than 1.

“Inflammatory reflex, through the activation of vagus nerve, controls the immune response to injury. Alterations of cardiac autonomic control we found in community-acquired pneumonia (CAP) patients could be seen as the final consequence of a complex interaction between systemic inflammation, that could trigger the inflammatory reflex, and hypoxia, that could trigger central and peripheral autonomic reflexes. Autonomic alterations in this population, characterized by a lower level of total variability and a reduction of LF components, could be partially related to the activation of the above mentioned excitatory different reflexes and these results were more evident in patients with severe CAP, compared to those without severe CAP on admission. In summary, CAP patients with a more severe disease on admission seem to show a loss of rhythmic sympathetic oscillation, with a predominant respiratory oscillation characterizing HRV. These findings were consistent across different evaluations of the severity of the disease on admission.” — Cardiovascular autonomic alterations in hospitalized patients with community-acquired pneumonia, Stefano Aliberti et al., published in Respiratory Research on August 04, 2016.

Another study has shown that an LF/HF ratio lower than 1 is associated with a 6-fold increase in the likelihood of pneumonia in immunocompromised patients. The risk of complications is also predicted by a low rMSSD value: if rMSSD is lower than 10, patients are 5 times more likely to have complications:

“90 patients with hip fractures admitted to Kongsberg hospital between 2008 and 2013 were invited to participate in the study.<…> One in three patients experienced at least one complication. Seven percent developed pneumonia, and two percent died before discharge. Due to low numbers of myocardial infarctions and stroke, these patients were not analysed separately, but were included in the analysis of overall complications. There were no differences in HRV parameters comparing patients with and without beta blockers. Patients with complications had significantly lower rMSSD and TP preoperatively (p = 0.043 and 0.03, respectively) compared to patients without. The likelihood ratio to develop complications in case of rMSSD < 10 was 4.9. Patients that experienced postoperative infections (pneumonia and urinary tract infection) had significantly lower VLF preoperatively, compared to patients without such complications (p = 0.04). In patients with postoperative urinary tract infections there was also decreased VLF (p = 0.02), and we found a tendency towards lowered VLF in patients with postoperative pneumonia (p = 0.091). There was a significant association between pneumonia and LF/HF<1 (p = 0.031). The likelihood ratio to develop pneumonia in case of LF/HF < 1 was 6.1.” — Decreases in heart rate variability are associated with postoperative complications in hip fracture patients, Gernot Ernst et al., published in PLOS One on July 25, 2017.

If you see such changes in the course of your COVID-19 infection, it’s a warning sign. Especially if the drop in heart rate variability measures is coupled with trouble breathing, persistent pain or pressure in the chest, confusion, or bluish lips or face.

You are in the hospital getting treatment for COVID-19 complications.

Now it’s the doctors’ responsibility to monitor your condition. But you can still use HRV as a science-based tool to track your recovery and the effectiveness of the treatments you get. Of course, that’s true only if you’re not already on a ventilator — in this case HRV readings will be altered by the machine and will no longer be valuable for health monitoring.

Doctors taking care of patients during the COVID-19 outbreak should also take a closer look at HRV. In a hospital that is over capacity, it’s difficult to make decisions about whose condition is truly critical and requires immediate assistance. Heart rate variability has been shown to be helpful in identifying patients who are in critical condition:

“A consecutive cohort of patients visiting the ED of a university teaching hospital who met the criteria of sepsis over a 6-month period were enrolled in this study. General demographics, vital signs, laboratory data, and Mortality in Emergency Department Sepsis score were obtained in the ED; the in-patient medical record was reviewed; and a series of continuous 10-minute electrocardiographic signals were recorded for off-line HRV analysis to assess the in-hospital mortality of the patients. One hundred thirty-two patients aged 27 to 86 years who met the inclusion criteria were enrolled. According to the in-hospital outcome, the patients were categorized into 2 groups: nonsurvivors (n = 10) and survivors (n = 122). The baseline HRV measures, including SDNN, TP, VLFP, LFP, and LFP/HFP ratio, of nonsurvivors were significantly lower, whereas the nHFP was significantly higher, than those of survivors. Multiple logistic regression model identified SDNN and nHFP as the significant independent variables in the prediction of in-hospital mortality for ED patients with sepsis. The receiver operating characteristic area for SDNN and nHFP in predicting the risk of death was 0.700 and 0.739, respectively. Heart rate variability measures, especially the SDNN and nHFP, may be used as valuable predictors of in-hospital mortality in patients with sepsis attending the ED.” — Heart rate variability measures as predictors of in-hospital mortality in ED patients with sepsis, Wei-Lung Chen et al., published in The American journal of emergency medicine in June 2008.

“Indexes of heart rate variability (HRV) appear to reflect severity and may have prognostic value in patients with acute exacerbation of chronic obstructive pulmonary disease (AECOPD). We hypothesized that AECOPD without adequate treatment response would demonstrate impaired cardiac autonomic regulation and changes in HRV after emergency department (ED) treatment. Patients with AECOPD requiring admission after ED treatment had a greater increase in HF% and greater decrease in LF/HF ratio compared to those discharged. Our study demonstrates patients with ΔHF% >7.1 or ΔLF/HF% ≦-0.39 require admission despite 24 h of ED treatment.” — Changes of heart rate variability predicting patients with acute exacerbation of chronic obstructive pulmonary disease requiring hospitalization after Emergency Department treatment, Tseng CY et al., published in Journal of the Chinese Medical Association in January 2018.

If you look at the data from the previously mentioned research, you will see that patients who are likely to get better in a week have a less severe drop in total power and VLF, and an LF/HF ratio just slightly below 1. More critical patients see their VLF drop more than 10-fold compared to healthy people, and their LF/HF ratio is around 0.78:

“A total of 48 patients (64 %) reached clinical stability before 7 days of hospital admission. In comparison to patients who reached CS within 7 days, those who reached CS after 7 days had significantly lower total power (p = 0.001) and lower VLF component (p = 0.003).” — Cardiovascular autonomic alterations in hospitalized patients with community-acquired pneumonia, Stefano Aliberti et al., published in Respiratory Research on August 04, 2016.

These changes in your HRV metrics can show if your current treatment is ineffective or if you require intensive care.

You haven’t been tested and are at home wondering if you have COVID-19.

Unfortunately, not everyone is able to get their hands on COVID-19 tests — there are no FDA-approved at-home kits yet, and medical professionals have to choose who gets tested. Young people with mild symptoms might just get a recommendation to stay at home for two weeks and monitor their symptoms in case they get worse. What can you do in this scenario?

Of course, HRV measurements won’t show if you have COVID-19 or another respiratory infection. But they can help you track the course of your illness and take action if you see a drastic drop in your scores.

TL;DR: You already have symptoms of a respiratory illness. How can measuring HRV help?

• HRV can’t help you differentiate between COVID-19 and other respiratory infections. If possible, you should get tested for SARS-CoV-2.
• If your symptoms are mild and you’re at home, you can use HRV to monitor your condition. A significant — 3 to 5-fold and more — drop in VLF and total power, coupled with difficulty breathing, is a sign that you’re not doing well.
• An important thing to monitor is your LF/HF ratio. If it drops lower than 1 and stays like that for a long time, this might mean that you have a higher risk of developing pneumonia.
• If you are already in the hospital with COVID-19 complications, it’s important to take your VLF power and LF/HF ratio into consideration. VLF power that has dropped 10-fold or more or an LH/HF ratio much lower than 1 — around 0.78 — are signs that your treatment isn’t working well.
• Doctors can also use SDNN to identify patients who need immediate assistance. People with significantly lower SDNN are less likely to survive, so they need to be admitted to intensive care first.
• If you are already on a ventilator, don’t use heart rate variability to track your condition and recovery — the readings will be altered by the machines.

How do I use HRV to track my recovery and overall well-being?

HRV isn’t a “one-size-fits-all” marker. There are some ranges for different metrics that are considered “normal,” but in fact your norm is strictly individual. That’s why it’s so important to track your HRV regularly: you’ll know your personal baseline for each metric and will see if something changes drastically.

Are you in good health? Great. It’s time to start tracking your HRV. Welltory’s smart algorithms need some time to calibrate your data and establish your personal baseline. It’s a good time for you to see what your measurements look like when you’re in good shape. Plus, it can curb your anxiety: if your HRV scores aren’t getting worse, you are probably not coming down with COVID-19.

Let’s recap. Here are the signs that you are fine:

• Time-domain metrics — SDNN and rMSSD — are stable or increasing
• For 300-beat measurements taken while lying down before bedtime each day, HF, LF and VLF are NOT dropping significantly over time
• LF/HF ratio is a little over 1 and stays that way

If your HRV metrics drop for several days in a row, you’ll know something’s up. Plus, you’ll be able to track your recovery if you do get sick — wait before your HRV scores stabilize before you go back to business as usual.

If you have been diagnosed with COVID-19 or are a researcher interested in studying disease progression patterns for COVID-19, we have launched an open data research project aimed at analyzing heart rate variability data from people sick with coronavirus here.

Join us! This is the first open data study for people diagnosed with COVID-19 that includes wearable and heart rate variability data.