Most people who wear a smartwatch think of it as a fitness gadget that counts steps and reminds them to get up. This picture is now seriously out of date. Today’s wearable devices can detect heart arrhythmias, monitor blood oxygen, alert caregivers when seizures occur, and even detect falls and call emergency services. The question is not whether these devices do impressive things. It’s whether they do these things reliably enough to matter to your health.
What wearables are really good
The strongest evidence of wearable health technology is continuous heart monitoring. Atrial fibrillation often causes no symptoms, so catching it passively while someone is just going about their day has real clinical value. The Apple Watch ECG feature received FDA approval in 2018, and since then multiple peer-reviewed studies have confirmed that it can detect AFib with clinically relevant accuracy when used correctly.
Epilepsy monitoring is another area where wearables have moved from consumer toys to medical devices. A number of companies are now developing Determining the seizure of the Apple Watch capabilities, making it a proper ecosystem of tools. EpiWatch, a division of Johns Hopkins Medicine, has received FDA 510(k) premarket clearance for its seizure detection platform, which reports 98% sensitivity for tonic-clonic seizures with a false alarm rate of just 0.08 per day. Apps like EpiCentr take a more comprehensive approach, connecting to the Apple Watch to monitor seizure patterns, abnormal heart rates and falls, and then sending GPS-targeted alerts to emergency contacts if something is detected. For a family managing epilepsy, having practical options is already a significant change from where things were a few years ago.
Sleep tracking has also come a long way. Devices like the Oura Ring have been evaluated in published studies against medical-grade actigraphy with reasonable agreement on total sleep time and other key metrics. This doesn’t make the loop a clinical sleep study, but it does mean the data is actually useful, not just decorative.
Where are the limits?
Understanding what wearables can’t do is just as important as what they can do. The difference between consumer and clinical devices is real, even if it is diminishing.
Two smartwatches measuring the same person’s heart rate variability may give different results, none of which match what the hospital monitor shows. The variability of measurement methods across devices complicates how to use this data, and manufacturers don’t always make their algorithms easy to compare or audit.
There are a few categories where even the most enthusiastic user should keep their expectations reasonable:
- Blood glucose monitoring in consumer wearables remains unreliable. The FDA issued a special safety communication in February 2024 advising people not to use smart watches or smart rings to measure blood glucose levels. Continuous glucose monitors like the Dexcom G7 work just as well, but they use a skin sensor rather than an optical wrist sensor.
- Most wearable seizure detection devices detect only one type of seizure. For example, EpiWatch is FDA-cleared specifically for tonic-clonic (grand mal) seizures. Currently, other types of passive seizures cannot be detected by the device, which is something that anyone considering it for a loved one should be aware of in advance.
- Consumer dressings are not a substitute for clinical evaluation. ECG readings from smartwatches can indicate something that needs to be investigated, but they cannot diagnose or rule out a condition. The doctor still needs to interpret the information in context.
Deregulation and why it matters
Not all “health features” on consumer devices pass the same level of testing. Wearable consumer devices marketed for general health and fitness may not need regulatory approval if they avoid specific medical claims. This means that some features marketed as “health monitoring” are closer to informed guesses than validated measurements.
The distinction is important in practice. When a device is cleared by the FDA, there is a specific process behind it that includes a pre-market proposal and a review of efficacy and safety risks before the product is sold. Algorithms used in refined devices are typically trained on large clinical datasets and tested against standards such as hospital EEG or Holter monitoring. Algorithms cannot be on unwieldy devices.
This creates a level view of what to look for before you buy:
- Cleared wearable medical devices, such as the Empatica Embrace (also cleared by the FDA for seizure detection) or the KardiaMobile 6L cardiac monitor, are approved for specific conditions and to produce data that doctors will actually use.
- Consumer smartwatches occupy a middle ground with some FDA-cleared features, such as the Apple Watch ECG or its sleep apnea detection in the Series 10. The cleared feature is confirmed; other health metrics on the same device may not.
- General health monitors provide trend data and broad patterns that are useful for motivation and awareness, but not appropriate for medical decision making.
Understanding your data
Even when the device is working as intended, the information is only helpful if you know how to read it. An increased heart rate during a stressful encounter is not a cardiac event. A sleep score of 68 is not diagnostic. Consumer devices often remove context in favor of clean panels, which can make normal variations alarming and true patterns easy to dismiss.
The most useful thing a wearable can do is help you discuss something with your doctor, not replace that conversation. If your device detects an irregular heart rhythm or a persistent drop in blood oxygen over several nights, this should be reported to your doctor. Acting on a random reading or ignoring a consistent reading are both ways to abuse technology.
The best wearable is the one that fits the actual needs of the person using it who knows what the numbers mean and what they don’t.
