Wearable Health Sensors: What They Can and Cannot Tell You

Regulatory agencies like the U.S. FDA distinguish between general wellness devices and medical devices. Most consumer smartwatches fall into the wellness category, meaning they are not required to meet the clinical accuracy standards that medical equipment must clear. Some specific features — such as certain ECG (electrocardiogram) functions — have received regulatory clearance for limited purposes, but that applies to one feature, not the device overall. Acting on wearable data as though it were a clinical diagnosis can lead to unnecessary anxiety or, worse, a false sense of reassurance.

Dedicated pulse oximeters are medical devices calibrated against arterial blood samples and validated for clinical use. Wrist-based SpO2 sensors use a different optical approach and are much more sensitive to movement, skin contact, and ambient light. Studies comparing the two have consistently found wider error margins in wearable devices. For casual wellness awareness — noticing a general pattern during sleep, for example — the feature can be informative. For anything medically significant, a dedicated device and professional interpretation are necessary. See which wearable health features are actually validated for more context on this distinction.

Pedometer-style step counting is one of the oldest and most refined wearable functions, and modern accelerometers do a reasonable job on straightforward walks. However, accuracy drops during activities like cycling, weightlifting, or pushing a stroller. Calorie burn estimates compound inaccuracies — they rely on step count, your entered body metrics, and assumed metabolic formulas that don't account for individual variation. Published research has found calorie estimates from consumer wearables can be off by 20–40% or more depending on activity type and the specific device.

Even the most feature-rich consumer wearable measures only a handful of metrics — typically heart rate, movement, skin temperature, and in some cases ECG or SpO2. Conditions involving blood pressure, cholesterol, blood glucose, organ function, or structural cardiac issues are entirely outside what these sensors can detect. Using the absence of an alert as health reassurance is a form of confirmation bias that the technology simply doesn't support. Regular check-ups with a healthcare provider remain the appropriate tool for comprehensive health screening.

There is no single physiological signal that directly measures "stress" or cleanly defines sleep stages. Wearables approximate both by combining data points — heart rate variability for stress, movement and heart rate for sleep staging — and feeding them through proprietary algorithms. The outputs are relative scores, not clinical measurements. Clinical sleep studies (polysomnography) use brain wave monitoring, eye movement tracking, and other inputs that a wrist device cannot replicate. Wearable sleep data can surface useful patterns over time, but individual nightly scores should not be over-interpreted.