The Science of Balance

Here is a summary of the science behind Feelmo, with sources. The fine details of the algorithms that make up your score are not public, but the ideas they rest on stand on published, peer-reviewed research.

This page is where we try to tell you, as honestly as we can, what Feelmo measures and why it measures it that way. It runs a little long, so feel free to skim only the headings that catch your eye.

Your heartbeat is not a metronome

A healthy heart does not beat in a fixed rhythm like a clock. The interval between one beat and the next (the R-R interval) is constantly fluctuating in tiny ways, in response to your breathing, posture, stress, and state of recovery. These fluctuations are what we call HRV (heart rate variability).

For example, your heart speeds up slightly when you breathe in, and eases off when you breathe out. This rhythmic fluctuation, synchronized with your breath, is one sign that the parasympathetic (vagal) nervous system that governs rest is working properly. Conversely, when tension or stress lingers, these fluctuations tend to shrink and flatten out.

Ever since a joint task force of the European Society of Cardiology and the North American Society of Pacing and Electrophysiology standardized how HRV is measured and physiologically interpreted (Task Force, 1996), HRV has been used in thousands of studies as a window for glimpsing autonomic nervous function from outside the body (Shaffer & Ginsberg, 2017).

What Feelmo looks at

HRV has many metrics, but Feelmo's Balance score is not a direct display of any single public metric. It summarizes multiple HRV-derived clues, measurement quality, and movement from your own baseline.

Time-domain views capture how much the beat-to-beat intervals vary over a period. Frequency-domain views separate faster, breathing-related fluctuations from slower patterns. Both are useful scientific lenses (Task Force, 1996; Shaffer & Ginsberg, 2017; Laborde et al., 2017), but no single lens should be treated as a direct answer to "how stressed am I?"

Feelmo uses these ideas as background, then compares your signals against your own baseline and the reliability of the measurement. The exact composition, weighting, and conversion method are Feelmo's proprietary analysis logic and are not disclosed publicly.

Why we compare against "your own baseline"

The absolute value of HRV varies greatly with age, sex, and constitution. Large-scale data spanning nine decades of life have shown that HRV declines with age (Umetani et al., 1998). In other words, the same number can mean something completely different from one person to the next in terms of "whether it is high or low for that individual."

That is why Feelmo looks not at how you compare with others, but at how you change relative to your own baseline. Whether you are higher or lower than the person next to you is not the point; how you compare with "your usual self" — that is the question that matters.

This is why the balance gauge on the home screen comes with an "aura" that represents your deviation from your own baseline. Rather than the score number itself, we care about gently showing "which way you are leaning right now, compared with your usual self."

Translating numbers into expressions

The balance score is a number from 0–100, but Feelmo does not put that number front and center. Instead, it translates the score into the expressions of six companions (Genki, Odayaka, Futsuu, Fuan, Panku, and Tsukare) and sets them quietly beside the score.

This is our way of not declaring a rise or fall in the number to be "good" or "bad." On a fluctuating day, "Fuan" (uneasy) or "Tsukare" (tired) may appear; on a settled day, "Odayaka" (calm) or "Genki" (lively) translate your current state into gentle, observational words. It is observation, never evaluation or diagnosis.

When to measure — why we value rest and sleep

Wrist optical sensors are convenient, but they have a weakness: they are vulnerable to body movement. Values taken while you are walking or swinging your arms are too noisy to rely on. So Feelmo places weight on conditions where the fluctuation can be read stably — on-demand measurement while seated still (measuring on the spot when you need to), and data taken during sleep.

There are several reasons we look closely at sleep.

• During sleep, the external factors that disturb HRV — exercise, food, posture, caffeine, and so on — are at their fewest, making it a time you can compare under near-identical conditions every day.
• During deep non-REM sleep, the parasympathetic nervous system has been shown to dominate (Trinder et al., 2001), so it is also a time when signs of recovery tend to show.
• The relationship between sleep and HRV has been organized in systematic reviews as well (Stein & Pu, 2012).

Stress and balance

The link between psychological stress and reduced HRV has been reported repeatedly in meta-analyses (Kim et al., 2018). Furthermore, a meta-analysis of neuroimaging studies has proposed a model in which the vagal pathway connecting the prefrontal cortex (the thinking brain) and the heart is what makes HRV a marker of stress and health (Thayer et al., 2012). For a closer look at this brain–heart link, see The Brain–Heart Connection.

The stress level and state displays on the home screen stand on this research background. That said, these are group-level "associations."

Why slow breathing works

The reason Feelmo's breathing sessions (the Breathing tab) guide you toward slow breathing (roughly six breaths per minute) is that breathing at this pace is known to resonate with the baroreflex (the blood-pressure-regulating reflex) and temporarily boost HRV significantly (Lehrer & Gevirtz, 2014).

At around six breaths per minute, the change in your heart rate as you inhale and exhale lines up exactly with the rhythm of the reflex that regulates blood pressure. When this "resonance" occurs, the fluctuation is greatly amplified. The psychophysiological effects of slow breathing have been reported in systematic reviews as well, where they are associated with heightened parasympathetic activity and an increased sense of relaxation (Zaccaro et al., 2018; Russo et al., 2017).

Both the "calm down" and the "lift up" sessions are gentle prompts built on this physiology of breathing.

On light and movement

There is a background to why the habit log in the Well-being tab handles sunlight and activity, too.

• Light is one of the strongest environmental factors affecting human circadian rhythms (the body clock), sleep, and mood (Blume et al., 2019). A record of morning sunlight is a small clue toward building a well-tuned body clock.
• Regular exercise has been reported in meta-analyses to be associated with increased HRV (Sandercock et al., 2005). Accumulated steps, too, become a foundation that supports your balance over the long run.

These are not quotas of what you "should" do, but soft clues for noticing the connection between your balance and your daily life.

What we can say about the score

The Balance score integrates multiple body signals, including HRV-derived clues, and translates them into a single score from 0–100.

• The specific structure and weighting of this integration are not disclosed, as a trade secret.
• The score is produced after also taking into account "how reliably we could actually measure that night or that session (the quality of measurement)."
• The validity of the analysis is continuously verified within a joint research project with Keio University.

Feelmo's principle is to watch over your balance rather than guessing your feelings. Rather than judging emotion itself, we aim to gently watch over how well your autonomic nervous system is balanced, on your own ruler.

What it means to measure at the wrist

The Apple Watch reads your pulse wave with an optical sensor and estimates R-R intervals. Under resting conditions, HRV derived from the Apple Watch has been validated to agree well with ECG-based measurement (Hernando et al., 2018).

Feelmo reads HRV, heart rate, sleep, and more from the Apple Watch / HealthKit on-device. Because accuracy drops during vigorous body movement, it makes sense, for this reason too, that we place weight on data taken during rest and sleep.

Checking it for yourself — the N-of-1 idea

Even when something is said to "work" on the average across a group, that does not necessarily mean it applies to you. So Feelmo's habit-impact analysis (Which habits help you, Feelmo Premium) is built on the N-of-1, within-person idea: it individually checks whether a habit such as breathing actually moves your own balance.

Confirming not "it works for everyone" but "it works for you," using your own data — it is a perspective that puts each individual at the center.

The fields of study behind Feelmo

Feelmo is designed by integrating not a single theory but multiple fields. Here we give an overview of "which disciplines it is built on." Note that the specific algorithms, computational methods, and parameters are Feelmo's own technology and are not disclosed on this page (we introduce only the field names and the ideas).

• Autonomic neuroscience and heart rate variability — reading the tendency of the balance between rest and tension from the wobble of your heartbeat (see the first half of this page for details).
• Affective science and psychophysiology — based on psychological frameworks that capture states along multiple axes, we express your current state in a few gentle categories (the six expressions).
• Signal processing and time-series analysis — carefully distinguishing noise from meaningful change in heart data taken during sleep and rest. We also consider "how reliably we could measure at that moment."
• Statistics and machine learning — taking the stance that a score should be captured not as a single point but as a "range," and treating the gap from "your own standard" statistically.
• Decision science — gradually learning "the way of tuning that suits this person, in this situation" from your data.
• Causal inference — incorporating ways of thinking that carefully distinguish "mere correlation" from "something that genuinely seems related."
• Personalization methodology (the N-of-1 idea) — looking at "what suits you," with each individual as the protagonist rather than the group average.
• Chronobiology — the same number can mean something different depending on the time of day or day of the week. We take daily and weekly rhythms into account.
• Behavioral science — we do not rush you to keep going. Rather than quotas or competition, we quietly put small accumulations into words.
• Balance and ethics — our wording avoids diagnosis and definitive claims, keeping things within the frame of tendencies and self-care. Learning and analysis are completed on-device, and ownership of the data belongs to you.

References

1. Task Force of the European Society of Cardiology and the North American Society of Pacing and Electrophysiology. Heart rate variability: standards of measurement, physiological interpretation and clinical use. Circulation. 1996;93(5):1043–1065.
2. Shaffer F, Ginsberg JP. An Overview of Heart Rate Variability Metrics and Norms. Frontiers in Public Health. 2017;5:258.
3. Laborde S, Mosley E, Thayer JF. Heart Rate Variability and Cardiac Vagal Tone in Psychophysiological Research – Recommendations for Experiment Planning, Data Analysis, and Data Reporting. Frontiers in Psychology. 2017;8:213.
4. Trinder J, Kleiman J, Carrington M, et al. Autonomic activity during human sleep as a function of time and sleep stage. Journal of Sleep Research. 2001;10(4):253–264.
5. Stein PK, Pu Y. Heart rate variability, sleep and sleep disorders. Sleep Medicine Reviews. 2012;16(1):47–66.
6. Kim HG, Cheon EJ, Bai DS, Lee YH, Koo BH. Stress and Heart Rate Variability: A Meta-Analysis and Review of the Literature. Psychiatry Investigation. 2018;15(3):235–245.
7. Thayer JF, Åhs F, Fredrikson M, Sollers JJ, Wager TD. A meta-analysis of heart rate variability and neuroimaging studies: implications for heart rate variability as a marker of stress and health. Neuroscience & Biobehavioral Reviews. 2012;36(2):747–756.
8. Lehrer PM, Gevirtz R. Heart rate variability biofeedback: how and why does it work? Frontiers in Psychology. 2014;5:756.
9. Zaccaro A, Piarulli A, Laurino M, et al. How Breath-Control Can Change Your Life: A Systematic Review on Psycho-Physiological Correlates of Slow Breathing. Frontiers in Human Neuroscience. 2018;12:353.
10. Russo MA, Santarelli DM, O'Rourke D. The physiological effects of slow breathing in the healthy human. Breathe (Sheffield). 2017;13(4):298–309.
11. Umetani K, Singer DH, McCraty R, Atkinson M. Twenty-four hour time domain heart rate variability and heart rate: relations to age and gender over nine decades. Journal of the American College of Cardiology. 1998;31(3):593–601.
12. Hernando D, Roca S, Sancho J, Alesanco Á, Bailón R. Validation of the Apple Watch for Heart Rate Variability Measurements during Relax and Mental Stress in Healthy Subjects. Sensors (Basel). 2018;18(8):2619.
13. Blume C, Garbazza C, Spitschan M. Effects of light on human circadian rhythms, sleep and mood. Somnologie. 2019;23(3):147–156.
14. Sandercock GR, Bromley PD, Brodie DA. Effects of exercise on heart rate variability: inferences from meta-analysis. Medicine & Science in Sports & Exercise. 2005;37(3):433–439.