Heart rate variability analysis can tell you a lot about how your body’s doing. To get accurate results, you need to stick to several simple rules. It’s essential because your heart responds to everything: your results can be affected not only by your position during the measurement, but also by the food you’ve eaten, the emotions you’ve experienced, the way you breathe, etc.
When you start using Welltory, it’s also important to train the self-learning algorithm, so that it can form your baseline. Regular morning measurements will do the job. They will help the algorithm understand your usual nervous system’s state before the effect of a breakfast, cigarette, coffee, or workout.
Why and how do I train the algorithm?
In Welltory, your HRV is analyzed by a self-learning algorithm. Its calculations are based on both standard heart rate variability metrics and your individual data. That’s why it’ll take some time for the algorithm to form your baseline and learn more about your body’s usual state.
To train the algorithm:
Take measurements at the same time every morning during a week or two (note that in Welltory, morning measurements are the ones taken between 5.00 AM and 12.00 PM).
Take measurements after waking up, but before working out, taking a shower, or eating breakfast. After you wake up, stay in bed for 5–10 minutes (don’t talk or check the news) and then take a measurement. You can use the bathroom or drink some water, too, but then you’ll need to lie down or sit down for 10–15 minutes before taking a measurement.
Always use the same position to take measurements (lying or sitting with your back against the back of the chair).
If you do a lot of sports, it’s best to take morning measurements in a sitting position. Sitting up after sleep, don’t forget to give your circulatory system several minutes to adjust before taking a measurement.
Training your algorithm usually takes a week or two. Once that happens, the insights you get will be more accurate. And if you’re taking measurements with the camera only, you’ll see a message in your feed once the algorithm is personalized.
How do I take accurate HRV measurements?
To check your measurement’s accuracy, tap the liquid message opening the detailed information — the accuracy score will be under the key insights.
You can also check the accuracy score if you open a measurement from Health Journal / Measurement History. If you are an iOS user, tap the chart sign in the upper right corner and go to Health Journal. If you are an Android user, tap Journal at the bottom of your screen. You’ll find the accuracy score right above your Stress, Energy, and Health scores in the measurement details.
When your measurement’s accuracy is high enough (95–100%), it means that the results on your screen closely reflect what’s going on with your body. To get high accuracy, you need to follow some simple rules.
Measurement accuracy includes two components:
what you do before and during the measurement (behavior quality)
how good the signal from your device is (technical quality)
Behavior quality includes your position during the measurement, any movements you’re making, the way you’re breathing.
To get high-accuracy measurements, follow these rules:
Lie down or sit down with your back against the back of the chair.
Before taking a measurement, wait for 10–15 minutes to let your heart rate get back to normal.
Try not to move. Any movement affects your heart rate.
Don’t try to control your breathing, intentionally taking deep or even breaths. The way you breathe affects your heart rate, so just breathe naturally.
Don’t talk. Talking changes your breathing rhythm.
Technical quality can be low if the signal is unclear or if the image is blurred (for camera measurements). If you’re using a heart rate monitor or Apple Watch, make sure they fit tightly. If you’re using your phone’s camera, the image quality may be affected by your finger’s position on the camera and flash or the light intensity in the room (it should be neither too bright nor too dark).
Find more details on taking measurements with compatible devices in our articles:
What can negatively affect the accuracy score?
High heart rate. If you’ve gone up a flight of stairs before taking a measurement or were actively moving during the measurement, your heart rate will be too high for a resting heart rate — and the accuracy score will decrease.
Unstable heart rate. If during a measurement your average heart rate changes significantly, your accuracy score will get worse. Say, you ran up a flight of stairs, waited for your heart rate to go down to 90 bpm, and started taking a measurement. And your usual resting heart rate is 60 bpm. During the measurement, your heart rate will gradually go down, which will reduce the accuracy score.
Moving, talking, or controlling your breathing during a measurement. When you move, talk, or control your breathing, it affects your heart rate. If there are abrupt changes in your heart rate, the algorithm may classify it as signal instability and filter out several beats, decreasing the accuracy score.
For camera measurements: incorrect finger placement. Our algorithm analyzes color changes in the image during a measurement. So, it’s important that your camera only captures your illuminated finger — that’s why you need to fully cover the camera with your finger and make sure the flash is illuminating the blood vessels. If the flash burns, hold your finger 2–3 mm away from it.
For camera measurements: small finger movements. Stability is essential for a good signal. If you move your finger or change its pressure on the camera, this will likely change the color of the image our algorithm analyzes. Several beats will be filtered out, reducing the accuracy score.
For camera measurements: bad lighting. If you’re in a room where the lighting is too bright or too dark, the camera may have trouble capturing your illuminated finger.
I follow all the rules, but my measurement accuracy is still low. What do I do?
One possible reason for low measurement accuracy is arrhythmia. If you have arrhythmia and are taking a measurement during an arrhythmia episode, the accuracy may be low. Arrhythmia affects your heart’s work and your heart rate variability. In this case, our algorithm can’t accurately analyze the state of your autonomic nervous system, or the balance of its sympathetic and parasympathetic parts, so it can't gauge your body’s state either.
Besides, our algorithm can’t yet distinguish between arrhythmia and bad signal. So, if you sometimes have arrhythmia episodes, it may take you several tries to take an accurate measurement.
If you take camera measurements, it can sometimes happen that you follow all the rules, but the accuracy is still low. In this case, try the following tips:
Try putting your finger so that its tip is on the camera and the finger itself is on the flash.
Then try putting it so that its tip is on the flash and the finger itself is on the camera.
Try holding your hand palm up with the phone on it and then try putting the phone face down on a table and putting your finger on top.
Try pressing your finger to the camera lightly, try pressing a little harder, and try not pressing at all.
If the flash gets too hot, you may slightly move your finger without even noticing it. If the flash burns, try holding your finger 2–3 mm away from it.
Try changing the lighting in the room (if it’s very bright, turn some of the lights off; if it’s dark, turn the lights on).
Experiment with the hand’s temperature: sometimes, when your hand is cold, it may negatively affect the measurement accuracy.
Try putting your finger on the camera before, during, and after the countdown. The autofocus technology or smoothing filters in smart cameras can sometimes be the problem.
Our users’ experience shows that there is no one universal method for taking accurate measurements. Try various options and monitor the accuracy — in most cases, it only takes a few tries to find the optimal position to get high-accuracy results.
Alternatively, you can take measurements with Apple Watch, Samsung HRM, or a heart rate monitor. You’ll find more information in our relevant articles: