Your smartwatch may not be as accurate as you think

Smartwatches can offer valuable health insights, but many figures they display are algorithmic predictions rather than direct readings from the body, according to a new study. These devices are frequently employed to monitor sleep, steps, heart rate, and other health indicators, yet their sophisticated metrics can be perplexing.

Assistant professor Adam Lepley from the University of Michigan School of Kinesiology and his team have devised a framework to assist users in comprehending what smartwatches measure, what they estimate, and how to interpret the data responsibly. Their findings are published in the journal Sensors. Lepley emphasizes that "not all smartwatch metrics should be interpreted the same way."

Some outputs closely correspond to the sensor's actual measurements, while many others are estimates formed by merging sensor data with proprietary algorithms, user-specific factors, and other presumptions. He advises against relying on these metrics at face value, stating that "in many instances, these gadgets are more apt for tracking trends over time rather than serving as precise laboratory measurements."

Smartwatches integrate data from various sensors, including optical sensors that use light to detect alterations in blood flow at the wrist, motion sensors and GPS that track movement, location, and additional signals. Algorithms then convert these signals into user-friendly measurements. The study concludes that smartwatch data are most beneficial for observing changes within the same individual over time.

A consistent change in resting heart rate, sleep patterns, or physical activity can carry more weight than a single unusual reading. Resting and stable heart rate, step count, and outdoor pace are generally more dependable than intricate estimations like calories burned, sleep stages, body composition, hydration, and recovery. The precision of these measurements can be influenced by factors such as movement, watch fit, temperature, sweat, skin tone, tattoos, and body composition.

Furthermore, the results may not be comparable across different brands, as manufacturers employ distinct sensors, definitions, and algorithms. The researchers executed a narrative review, utilizing topic-focused searches in PubMed, SPORTDiscus, and Google Scholar up to June 2026, and surveyed reference lists, technical and regulatory documents, and professional guidance.