Patients arriving at sleep clinics with wearable sleep data is now the norm rather than the exception. Apple Watch, Oura, Whoop, Garmin, Fitbit — the data formats and metrics vary, but the conversation that follows is similar. The data sometimes adds useful clinical signal. Often it adds noise. Knowing which is which has become part of the clinical skill set in 2026.

Where the data is clinically useful

The areas where consumer wearable sleep data adds meaningful clinical signal are narrower than the marketing suggests. Total sleep time tracking over weeks or months is the most reliable. The trend data — averaging across many nights — is more useful than the individual night reports.

Heart rate variability data during sleep can flag autonomic dysregulation that prompts further investigation. The data is not diagnostic but it can be a useful screening signal.

Sleep timing — when the patient is actually getting into bed and getting out of bed — is more reliable than self-report and is clinically useful for circadian conversations.

Where the data is misleading

Sleep stage classification by consumer wearables is unreliable enough that it should not be used clinically. The wearables call something “REM sleep” but the classification accuracy against laboratory polysomnography is poor. Patients arriving worried about low REM percentages are usually worried about a number that does not mean what they think it means.

Sleep score composite metrics are similarly unhelpful clinically. The scores combine various inputs in opaque ways and are designed for consumer engagement rather than clinical interpretation. A poor score does not mean a poor sleep night in any sense the clinical team can act on.

Snore detection and breathing irregularity detection on consumer wearables is improving but still produces false positive and false negative results often enough that the data should not substitute for sleep study results.

The conversation that works

Patients are emotionally attached to their wearable data. Dismissing it produces resistance. Engaging with it carefully, while explaining what it can and cannot tell us, produces better conversations.

The framing I find useful is to treat the wearable data as a long-term trend monitor rather than a diagnostic instrument. The patient who has six months of data showing a steady decline in total sleep time has produced useful information. The patient who is worried about last night’s poor sleep score has produced anxiety rather than information.

When to order a sleep study

Wearable data does not replace a sleep study when there is clinical suspicion of obstructive sleep apnoea, central sleep apnoea, narcolepsy, or other parasomnias that require polysomnographic confirmation. The wearable data can prompt investigation but it cannot conclude it.

The patients who are most resistant to sleep studies often have the most concerning wearable data patterns. Translating the wearable findings into the case for a sleep study is part of the consultation work that takes time but pays off.

A note on the home sleep study category

Home sleep studies, which are clinical-grade rather than consumer-grade, occupy a different space and do produce data that can substitute for laboratory sleep studies in many cases. The category continues to mature and is increasingly the appropriate first investigation for suspected OSA.

The line between consumer wearables and home sleep studies is clinically important. Patients sometimes conflate them. Clarifying the difference is worth doing early in the conversation.