I'm curious about the underlying tech behind smartwatches' health monitoring. Specifically, how do they use optical sensors to detect heart rate, and what role do accelerometers and gyroscopes play in tracking steps and movement? Also, how are the raw signals processed into the metrics we see on the screen? Any clear explanations or resources would help. What do you think are the biggest challenges in making these measurements accurate?
How do smartwatches measure heart rate and activity?
👁️ 73 views💬 2 replies❤️ 0 likes
2 Replies
Optical heart rate sensors in smartwatches primarily use PPG (photoplethysmography) technology. On the back of the case, there’s an LED (usually green, sometimes infrared) and a photodetector. With each heartbeat, blood changes the skin’s optical density, and the photodetector picks up tiny fluctuations in reflected light. The raw analog signal is converted via an ADC, then processed in firmware with several filters: a high-pass filter (to remove slow drift), a low-pass filter (to suppress noise above 4 Hz), and often an adaptive filter that uses accelerometer data to correct motion artifacts. After cleaning the signal, the interval between peaks (R-R) is calculated, and the heart rate in beats per minute is derived.
The accelerometer and gyroscope in the watch serve dual purposes—step counting and advanced activity analysis. A 3-axis accelerometer measures acceleration along X, Y, and Z, while the gyroscope tracks angular velocity, helping distinguish between normal steps, running, or smooth body movements. The step detection algorithm typically uses an acceleration threshold and checks if a sequence of impulses matches a characteristic gait pattern. The gyroscope refines body orientation (sitting, standing, lying down) and eliminates false triggers when the watch is in hand but the user isn’t walking (e.g., raising the arm to check a notification).
The biggest accuracy challenges come from mechanical artifacts and individual skin differences. During intense workouts, arm movement creates strong noise that can "drown out" the PPG signal, so without a robust cross-correlation filter, most readings become unreliable. Additionally, skin pigmentation, tattoos, epidermal thickness, and temperature affect light absorption, requiring the model to be calibrated for each user. Another hurdle is battery life: more frequent measurements improve quality but drain the battery quickly, forcing developers to balance sampling frequency with power consumption.
Thanks for the question – the smartwatch’s PPG sensor uses a green LED and photodiode to detect blood-volume changes, while the accelerometer and gyroscope count steps and infer activity by measuring linear and angular motion; the raw waveforms are filtered, peak-detected, and combined with motion data in firmware to output BPM and calorie estimates. Do you find the heart-rate readings less reliable during intense workouts, or is the motion-artifact filtering holding up for you?