​HRV and sleep duration was collected with an Oura ring. Training data was collected with Garmin GPS watch and Polar sensor (OH1 or Verity Sense). All data was read in HRV4Training and analyzed in HRV4Training Pro in the long term. For more information about aerobic efficiency, see this article.

​As shown in Box1, this period was a training phase for a 24h race held in late May. The purpose of training is to increase volume (km). Training included long-distance running (40-80 km/session) and frequent jogging (about 15km/session). I checked HRV frequently and was careful not to increase the training load when the HRV was trending down. I felt that the approach may contribute to a stable HRV (and a slight increase in HRV). Although I was unable to break my own record from eight years before (181.690km), I was able to perform to a high standard in the race (171.760km). You can read about my approach to the race in this peer-reviewed paper.

Congratulations to Dan Plews for taking the win at Ironman New Zealand

Here are his last weeks of HRV data, showing:

• an ideal response to training
• minor sickness pre-race, recovered quickly
• a large, sustained drop post-race

The visualization below, including daily HRV, normal range and baseline, as well as the detected trend (which combines HRV, heart rate and coefficient of variation), is part of HRV4Training Pro, which you can try here if you are using the app: https://www.hrv4t.com

all the best for your recovery, Dan

More and more wearables have started capturing Heart Rate Variability (HRV) data overnight and combining it with other parameters (e.g. heart rate, sleep data, physical activity) to provide readiness or recovery advice to the user.

Despite some inconsistencies over the past years, as of the end of 2022, Oura, Whoop, and Garmin all work in a very similar way when it comes to HRV measurement. However, the way the data is used in all of these tools when building readiness or recovery scores or when providing advice to the user is often problematic and inconsistent.

Naive interpretations (higher is better), lack of a normal range (what’s a meaningful change?), and confounding your physiological response with your behavior, are common issues that limit the utility of data collected with wearables.
In my latest blog, I cover data analysis and interpretation to provide you with some useful tips and tools that should allow you to make the most of the collected data and ignore inaccurate interpretations provided by most tools out there.
Thank you for reading.

The subjective score in HRV4Training aims at capturing how you feel in response to training, and combines: perceived sleep quality, muscle soreness, motivation to train, perceived performance during training.

In the app, when providing daily advice (color-coding and message) in HRV4Training we combine your physiology and your subjective feel (outputs) . 

However, we do not use or include your behavior, for example your activity / training (input). This is a key difference from what you get in terms of readiness or recovery scores in most wearables. Why is that?

The whole point of assessing your state, either objectively via heart rate variability (HRV) or subjectively by feel, is to determine how you responded to your given circumstances. You already know the input (behavior) and are assessing the output (physiology or feel).

In other words, if I train hard or more for a few days, I want to assess how I responded (output). Including activity (input) in my assessment would mean penalizing me regardless of my body's response. For athletes (of any level), this method is particularly ineffective: it hides information.

If you train, there is no point looking at readiness or recovery scores to assess how you are responding to a given training stimulus as these scores confound your response with your behavior. Is the score low because I responded poorly, or just because I did more?

The subjective score in HRV4Training is not impacted by these limitations, as it relfects your subjective input. When combined with your physiological response (heart rate, HRV), it can give you a more comprehensive picture of your response, and help you make meaningful adjustments to your plans. ​

Citing Wikipedia: ​"Correlation refers to any of a broad class of statistical relationships involving dependence. Familiar examples of dependent phenomena include the correlation between the physical statures of parents and their offspring, and the correlation between the demand for a product and its price. Correlations are useful because they can indicate a predictive relationship that can be exploited in practice."

In other words, looking at correlations can help us to pinpoint which parameters have a stronger impact on our physiology, and potentially make adjustments (e.g. if there is a strong negative correlation between work stress and HRV, maybe we should try to reduce work stress). 

In the latest update of HRV4Training Pro (try it here), we have added your subjective score to the list of parameters you can pick in the Tags view.

You can see an example below, where the third plot shows my subjective score in the past few months, which is quite in agreement with my HRV data. In particular, we have first a slow increase over time as I am progressing well with training, eventually racing a half marathon PR, before an acute drop with sickness. 

We hope you will find this feature useful.

Check out HRV4Training Pro at this link or use code SCIENCE for 20% off any plan.

​Enjoy.

The swiping between views will be enabled if you select rMSSD under HRV View in Settings, and have a Pro plan.

Pro enables also the normal range under Baseline (see previous image), and gives you access to the web platform you can try Pro for free here: https://hrv4t.com.

With HRV4Training Pro, you get:

• access to the web platform to analyze data in the long term
• rMSSD view on the homepage
• normal range in the Baseline view
• you support this independent business and make sure we are still here next year

Thank you and enjoy 

In case you would like to use your Oura ring to measure your HRV in the morning, as opposed to the night, you can do so following the procedure below.

If you are wondering why you would want to measure HRV in the morning instead of the night, here you can find some useful pointers about the differences between these two measurement times. 

From the Oura app, tap the plus icon, and start an Unguided session:

Make sure to set the unguided session duration to 3 minutes, otherwise the ring will not report HRV data, but only heart rate:

At this point, select Manual Input from Settings in HRV4Training, so that you can enter your heart rate and HRV in the HRV4Training questionnaire, instead of measuring via the app.

You can learn more about Manual Input, here.

​Enjoy.

here is an example of how we would recommend holding the phone when using three (left) or two (right) cameras.

In case you cannot cover both the flash and the camera at the same time, give priority to the camera, as it is sufficient to be nearby the flash.

Enjoy the update!

We have released Manual Input in HRV4Training, a feature which allows you to enter manually your heart rate and HRV data in the morning, as part of the questionnaire. 

​Below we cover the reasoning behind this feature, and how you can use it. 

More and more devices are providing night HRV data based on optical measurements, with good accuracy. For example, in our recent tests comparing full night electrocardiography (ECG) against the Oura ring, Whoop band and Garmin Forerunner 955, day to day differences in HRV were very similar. 

So far we have provided a direct link only to Oura. However, unfortunately, this link has resulted in very unreliable data connection due to Oura's API having quite a few problems. Additionally, it seems the API has been shut down for many users, probably due to the new subscription model or using the older ring.

Similarly, other wearables do not have APIs (e.g. Whoop) or we simply do not integrate with them (e.g. Garmin, Fitbit, etc.). 

Since we are talking about just two numbers here (resting heart rate and HRV, I have covered elsewhere how the full night average is the only meaningful parameter to use when measuring during the night), Manual Input makes it really simple to avoid all issues above, and allows you to enter your data as part of the morning questionnaire, regardless of the wearable you use. 

Simply put, HRV4Training is the only platform that provides you with an analysis of your physiology that matches how this data is used in state of the art research and applied practice.

This means analyzing your resting physiology with respect to your normal range, and providing you with feedback regarding your acute (daily) and chronic (weekly) physiological state, in response to the various stressors you face. 

While in most tools you can look at your HRV numbers, these numbers are not contextualized with respect to other events (e.g. annotations you report in our questionnaire), or even with respect to your historical data.

Is today's reduction in HRV meaningful or just part of normal day to day variability? Entering your data in HRV4Training allows you to answer this question using published methods, which aim at effectively assessing your physiological response.

In the homescreen of the app you can easily see for example your daily scores with respect to your weekly baselines and normal ranges. In particular, the normal ranges are built using the previous 2 months of data, and allow you to quickly understand your current physiological response. 

You can learn more about why you should focus on your actual physiology as we do in HRV4Training, as opposed to made up scores provided in wearables, in this blog and this podcast. 

Wearables use slightly different methods and sensing locations to measure your HRV. These differences make it so that the data is not interchangeable, but the relative changes, which are the only aspect that really matters, are. Hence, it is important that if you wear a wearable, you try to use always the same device and place it on the same arm, finger, etc, since the relative position of the sensor with respect to the heart, can cause differences in pulse rate variability.

If you use a Whoop band, enter your night resting heart rate and HRV as found in the app. If you use an Oura ring, enter your average night heart rate (not what they report as resting heart rate, which is the lowest, enter the average instead) and your average HRV. If you use a Garmin, enter your night average HRV and your resting heart rate (Garmin uses the 30 minutes of the night in which it was the lowest).   

If you do not have a wearable and you have been measuring your physiology in the morning daily, you probably do not need a wearable and can save some money. 

If you are struggling with compliance for your morning measurement and would like to collect data passively, a wearable can help.

In any case, check out this blog to better understand some of the important differences between morning and night measurements of resting physiology.

I hope you'll find the update useful, thank you

When providing daily advice (color-coding and message) in HRV4Training we combine your physiology and your subjective feel (outputs) . However, we do not use or include your behavior, for example your activity / training (input).

this is a key difference from what you get in terms of readiness or recovery scores in most wearables. Why is that?

The whole point of assessing your state, either objectively via heart rate variability (HRV) or subjectively by feel, is to determine how you responded to your given circumstances. You already know the input (behavior) and are assessing the output (physiology or feel).

In other words, if I train hard or more for a few days, I want to assess how I responded (output). Including activity (input) in my assessment would mean penalizing me regardless of my body's response. For athletes (of any level), this method is particularly ineffective: it hides information.

If you train, there is absolutely no point looking at readiness or recovery scores to assess how you are responding to a given training stimulus as these scores confound your response with your behavior. Is the score low because I responded poorly, or just because I did more?

This approach not only provides you with poor information about your actual response, but fools you to believe the tool works. You go hard or do more, and they tell you you need to recover. In fact, you might be doing very well and be ready for another big training block.

This is not to say that your behavior does not matter: it is key context you can use to understand what could be driving changes. However, it should not be used to determine your response (output). You want to learn about the output of the system (physiological or subjective response) given the input (behavior and other).

There are many nuances that are worth understanding a bit better if we want to make good use of available technology. Hopefully, this explains a bit why it is worth assessing your physiology and feel, while you can ignore most (all?) made-up scores. 

In the context of analyzing relative changes over time, for example to identify periods of higher stress, there are two important trade-offs to consider when it comes to resting heart rate and HRV:

• using a short time window (e.g. a few weeks) would be too reactive: a period of poor health would become your normal quickly. We want to prevent this from happening, as even long periods of poor health (due to sickness or injury for example) should not be considered your normal state
• using a long time window we would be stuck in seasonality, and not change dynamically as our physiology does over a year. This is also something that we need to avoid, as we need to compare your physiology today with relevant and current historical data, and not with your data many months ago

In the scientific literature, for practical reasons, often one month is used to determine the normal range. However, we need to realize that scientific studies typically face obvious constraints (e.g. time and budget) and as such, might be trying to shorten the time required to capture an individual's normal range.

I would like to argue that this is too short and ineffective to capture longer-term decouplings between baseline (weekly average) and your normal range. Let's look at an example:

In the data above, towards the end of the 3 months, we can see a few bad weeks where HRV is quite suppressed.

Let's look at the first graph first. If we were to build the normal range using only a month of data, the normal range would change too quickly: it would always include the baseline (blue line) despite a very large change in daily scores. In other words, if we use short windows, we are almost always within normal range.

Now moving to the second graph. When using 60 days of data to build the normal range, we can see how the normal range decouples more effectively from the baseline and daily values. In this case, we can clearly see that we are in a negative phase of suppressed daily and baseline HRV, with respect to our normal range.

In HRV4Training we use 60 days of data for these reasons. Obviously, ​there are always trade-offs to make and no choice is perfect, but in our experience, 2 months is an ideal time frame when looking at HRV data: you are not too reactive and can capture acute drops, you don't get stuck in very old data and seasonal changes.

You can try Pro for free by logging in here. Once you are logged in, the Baseline page in the app will also show your normal range, as shown below. Use code SCIENCE for 15% off.

​Enjoy

Blog post by Marco Altini

We have just released a new version of HRV4Training for iPhone and Android, which deploys our latest VO2max model. 

In particular, this model better accounts for hilly terrain and provides more accurate VO2max estimates for runners that train on hills.

Normally, the relationship between running speed and heart rate is used to estimate VO2max, as we cover in more detail here. 
Intuitively, a lower heart rate at the same speed, means that you are getting fitter. However, this relationship falls apart when we run on trails, or include much elevation gain in our runs. Below is an example of my own data, where you can see a few things:

• until the end of 2017, I lived in a hilly city (San Francisco). Then, I moved to a flat one (Amsterdam)
• the only real change in VO2max was in 2016-17, when I started with polarized training (details here)

In the last part of the graph you can see how my interest shifted toward trail and ultrarunning (very few flat runs). As a result, VO2max estimates show a large decrease. In fact, little change is present in speed over heart rate when looking only at these flat runs (yellow line).

This makes it hard to track progress in aerobic efficiency or VO2max (or whatever you want to call how your heart rate changes at a given speed). 

With the newly released model, we are able to better account for these changes, and provide a more accurate estimate. 

To build this, we modeled the difference in estimated VO2max for the same person when running flat vs hilly, in relation to the average grade of the runs (N = 10 000).

There are of course still limitations, for example no knowledge of how technical is the terrain, but it should be an improvement.

Below are results showing the original estimate (which was a match with testing in the lab), then the poor performance of the model on hilly runs, and the updated results today, for me (right) and Alessandra (left).

​We both had an accurate estimate when living in The Netherlands, then got a reduction in VO2max simply because of the different efficiency (relationship between pace and heart rate) when running on hills in Italy, and then finally you can see the prediction of the new model, which is close to the original one.

With the HRV Logger you can take measurements before and after exercise, and run the same comparisons shown below, directly in the app (the Compare tab is available only on iPhone).

Make sure to configure the app as below:

• 30" windows
• 25% artifact correction

I would like to add a note about artifacts here and how the Compare view allows you to filter them out even when the RR intervals timeseries is impacted. Post-exercise, these days I have many ectopic beats. You can see below that some of them remain even after artifact correction (the spikes in the second recording, right end side).

If your HRV stays suppressed after 24 hours since your workout, most likely:

• training was too hard for your current fitness level
• ​the training stimulus was novel
• non-training related stressors played a role

These are key points highlighted by Andrew Flatt in his research. 

When it comes to recovery, HRV is one piece of the puzzle. Having recovered quickly from a nervous system point of view means you have the capacity to assimilate the stimulus. Additionally, a stable HRV highlights how you are also dealing well with non-training related stressors (even more important).

Yet, there should be no surprise if your daily HRV does not correlate with muscle soreness or feel (whatever that means) or performance. HRV is your response: keep it stable.

If today you are resting and your HRV looks good, you are in an ideal situation. 

For any feedback on this blog, feel free to reach me here.

We've updated our Ultimate Guide to Heart Rate Variability (HRV), part 1:

• short background
• similarities and differences between morning and night data
• best practices for your measurements
• HRV metrics
• a note on normal values

You can find it here.

We hope you will find it useful, thank you for reading!