Wearable devices use a handful of sensor technologies to estimate physiology: photoplethysmography for heart rate and blood oxygen, accelerometers for steps and activity, and single-lead electrodes for on-demand electrocardiography. Across peer-reviewed research and regulatory review, a consistent theme emerges: these sensors can be clinically useful for specific measurements and contexts, but they are not uniformly accurate across all conditions or populations.
How accurate are heart-rate and ECG sensors?
Photoplethysmography, or PPG, measures pulse by detecting blood-volume changes with green or infrared light. Studies that compared wrist-worn PPG against chest-strap electrocardiography report that PPG gives reliable heart-rate estimates at rest and during steady-state activity but degrades during intense motion or irregular rhythms. Mintu P. Turakhia at Stanford University led large-scale clinical work on smartwatch irregular-rhythm detection that demonstrated wearables can flag potential atrial fibrillation for follow-up, a result that helped support regulatory pathways. The U.S. Food and Drug Administration has cleared specific smartwatch ECG and arrhythmia-notification features as medical devices, which indicates those functions met standards for safety and intended use. Even so, single-lead ECG from a wrist device is not a replacement for a 12-lead clinical ECG when comprehensive cardiac assessment is required.
Sources of error, equity, and real-world consequences
Accuracy is affected by motion artifact, ambient light, sensor placement, and physiology such as body habitus and skin tone. Matthew W. Sjoding at the University of Michigan and colleagues documented that pulse oximetry—related optical technology used in some wearables—can be less accurate in people with darker skin, a finding that prompted additional scrutiny by clinicians and regulators. Environmental and territorial factors also matter: devices may perform differently in hot, humid conditions or at high altitude, and populations with limited access to updated models or repair services face inequities in both performance and benefit. Eric J. Topol at Scripps Research has emphasized that digital health tools can extend care but also risk widening disparities unless deployment is accompanied by clinical validation across diverse populations.
Beyond heart rate and rhythm, estimates that many wearables provide—calories burned, sleep stages, and exact blood-oxygen saturation—show wider variability in independent studies. Those metrics are often best used for personal trend tracking rather than single-instance medical decision-making. False positives can cause anxiety and unnecessary medical visits; false negatives can delay needed care.
Clinically, the most defensible use of wearables today is as screening and monitoring adjuncts: they identify signals that merit clinical evaluation, facilitate remote monitoring when combined with professional oversight, and support behavior change through feedback. For definitive diagnosis or treatment decisions, rely on clinical-grade testing and physician interpretation.