What photoplethysmography actually measures, how a fingertip camera scan turns light into a heartbeat, what the resulting numbers mean, and where the real limits of phone-based measurement are.
In this article
What PPG actually is
Photoplethysmography (PPG) is an optical technique for detecting blood volume changes in the microvascular bed of tissue. Shine light into skin, measure how much of it comes back, and you'll see that reflected (or transmitted) light intensity rises and falls in step with each heartbeat — because with every cardiac cycle, a fresh pulse of arterial blood briefly changes how much light the tissue absorbs. That rhythmic optical signal is the PPG waveform.
It's not new. Clinical PPG sensors have been standard in hospitals for decades — it's the technology behind the fingertip pulse oximeter clipped onto a patient's finger in an ICU or operating room, most commonly used there to estimate blood oxygen saturation alongside heart rate. What's changed is the hardware needed to do it: a dedicated LED-and-photodetector sensor is no longer required, because a phone's rear camera and flash can approximate the same setup.
How a phone camera harvests a pulse signal
When you place a fingertip directly over a phone's rear camera lens with the flash on, the flash LED illuminates the tissue and the camera sensor records the light that reflects back, frame by frame, effectively turning the camera into a photodetector. As blood volume in the fingertip's capillaries pulses with each heartbeat, the amount of light absorbed by the tissue changes very slightly — and that shows up as a tiny brightness fluctuation across successive video frames.
That raw signal has two components, and separating them is the entire game:
- A large, slow-moving DC component — the baseline absorption from skin, tissue, bone, and venous blood, which changes gradually with breathing and sensor pressure but carries no beat-to-beat information.
- A small AC component riding on top of it — typically only around 1–2% of the total signal — which is the actual pulsatile arterial blood volume change caused by each heartbeat.
Signal processing (bandpass filtering, detrending, peak detection) isolates that small AC component from the much larger DC baseline and noise, producing a clean pulse waveform with a distinct peak for every heartbeat. The time between consecutive peaks is the beat-to-beat interval that every downstream metric — heart rate, HRV, respiratory rate estimates — is built from.
What cardiac parameters PPG can derive
Heart rate
The most direct output: count the peaks over a window of time (or measure the interval between consecutive peaks and invert it), and you have beats per minute. This is the most robust PPG measurement — well-validated against ECG in resting, low-motion conditions.
Heart rate variability (more precisely, pulse rate variability)
HRV describes how much the interval between consecutive heartbeats varies, rather than the average rate itself — a healthy autonomic nervous system continuously adjusts that interval in response to breathing, blood pressure, and physical and emotional load. From a PPG waveform, the metric being measured is technically pulse rate variability (PRV) — beat-to-beat interval variability derived from arterial pulse timing — rather than the R-R interval variability that clinical HRV is classically defined from on an ECG. The two agree closely under resting, still conditions, but can diverge during movement or physiological stress, which is a real limitation of camera-based measurement worth being upfront about (see Limitations).
The commonly reported sub-metrics:
- RMSSD — root mean square of successive beat-to-beat interval differences; the standard short-term, time-domain marker most sensitive to fast, breath-linked (parasympathetic) fluctuation.
- SDNN — the standard deviation of all beat-to-beat intervals in a recording; reflects overall variability from every source acting on heart rate over that window, not one specific branch of the nervous system.
- Frequency-domain LF and HF power — the same beat-to-beat interval series decomposed into low-frequency (~0.04–0.15 Hz) and high-frequency (~0.15–0.4 Hz) bands. HF power tracks closely with breathing-linked (parasympathetic) modulation; LF power reflects a mix of influences that is still debated (see the callout below).
Respiratory rate
Breathing subtly modulates both heart rate and pulse amplitude (respiratory sinus arrhythmia), so an approximate breathing rate can be extracted from the same PPG recording without a separate sensor — a genuinely useful byproduct, though a coarser estimate than a dedicated respiration sensor.
Pulse waveform shape
Beyond timing, the shape of each individual pulse — how sharply it rises, whether there's a visible secondary reflection wave — carries information related to arterial stiffness and vascular tone in research settings. This is the most exploratory of the parameters listed here and the one we treat most cautiously in our own apps.
Established: that beat-to-beat interval variability reflects autonomic nervous system activity, and that RMSSD and HF power track parasympathetic (vagal) tone reasonably well.
Still debated: the long-used LF/HF ratio as a clean marker of "sympathovagal balance." Several physiologists have argued this specific interpretation oversimplifies what the LF band actually reflects. We report LF/HF because it's part of the conventional metric set, but we do not present it to you as a settled, precise dial for stress versus calm.
What's "normal," and why that's a loaded question
Resting heart rate has a relatively well-agreed general reference range. HRV does not — it varies enormously by age, sex, fitness level, genetics, and even measurement conditions, to the point that a single population-wide "normal HRV" number is not clinically meaningful. What follows are general reference points from the literature, not personal targets.
| Parameter | General adult reference | Key caveat |
|---|---|---|
| Resting heart rate | ~60–100 bpm | Well-trained endurance athletes commonly sit lower (40s–50s) as a normal adaptation, not a concern. |
| RMSSD (short-term) | Roughly tens of milliseconds, wide individual range | Declines with age; healthy individuals of the same age can still differ several-fold. |
| SDNN (short-term) | Wide individual range, generally higher than RMSSD | Very sensitive to recording length and conditions — not comparable across differently measured sessions. |
| Respiratory rate | ~12–20 breaths/min at rest | Estimated indirectly from the PPG signal, so it's coarser than a dedicated respiration sensor. |
The practical takeaway: don't compare your HRV number to a friend's, a headline figure, or a leaderboard. Compare it to your own baseline over time.
Reading your own numbers for wellness, not diagnosis
Because normal ranges are so individual, the useful signal in day-to-day self-tracking is almost always the trend, not any single reading:
- Establish a baseline first. Several measurements over 1–2 weeks, at a similar time of day and in similar conditions, before drawing any conclusions.
- Watch direction, not absolute value. A sustained drop in your own HRV or a sustained rise in your own resting heart rate over days to weeks is more informative than any one morning's number.
- Expect noise. Sleep, hydration, caffeine, alcohol, illness, and recent exercise all shift these numbers day to day.
- Use it as a prompt to look closer, not a verdict. A persistent, unexplained change is a reasonable reason to mention it to a clinician — it is not, on its own, a diagnosis of anything.
Why this is worth tracking at all
The case for phone-based physiological self-tracking isn't that it replaces clinical measurement — it's that it fills a gap clinical measurement structurally can't: continuity between visits. A cardiology appointment captures one heart rate, on one day, in a clinical setting that itself often elevates it ("white coat" effect). A week of your own resting-state readings, taken the same way each time, can surface a gradual trend long before it would show up as a single abnormal value in an office visit.
Where phone-based PPG genuinely falls short
- Motion artifact. Even small finger movement or pressure changes during a scan can distort the signal far more than it would in a fixed clinical sensor.
- Skin tone and perfusion. Signal quality from optical sensors can vary with skin pigmentation, skin temperature, and peripheral circulation — a documented issue across the pulse-oximetry and PPG literature generally.
- Lighting and camera hardware variability. Different phone models' camera and flash characteristics aren't identical.
- PRV is not identical to ECG-derived HRV. The two track well at rest but can diverge under motion or stress.
- Not a diagnostic-grade instrument. None of this is validated against clinical-grade ECG or approved as a medical device.
References
- Allen J. "Photoplethysmography and its application in clinical physiological measurement." Physiological Measurement, 2007.
- 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.
- Shaffer F, Ginsberg JP. "An overview of heart rate variability metrics and norms." Frontiers in Public Health, 2017.
- Nunan D, Sandercock GRH, Brodie DA. "A quantitative systematic review of normal values for short-term heart rate variability in healthy adults." Pacing and Clinical Electrophysiology, 2010.
- Schäfer A, Vagedes J. "How accurate is pulse rate variability as an estimate of heart rate variability?" International Journal of Cardiology, 2013.
- Billman GE. "The LF/HF ratio does not accurately measure cardiac sympatho-vagal balance." Frontiers in Physiology, 2013.
- Castaneda D, Esparza A, Ghamari M, Soltanpur C, Nazeran H. "A review on wearable photoplethysmography sensors and their potential future applications in health care." International Journal of Biosensors & Bioelectronics, 2018.