How to Evaluate a Health Device: A Verified-Specification Checklist

What this checklist is

This is a way to check a device claim yourself, using specifications and evidence you can verify. It ranks no brands or models and names none as best. A device claim is only as strong as the evidence for that exact model, that exact function and that kind of person.

Evidence labels used: established, mixed, insufficient evidence, marketing claim. See How We Label Evidence in Biohacking. For reading the studies themselves see How to Read a Wearable Validation Study; for marketing language see How to Read Health Product Marketing Claims.

The checklist

1. Write the claim in one sentence. For example "measures X" or "improves Y". If you cannot, the claim is vague.

2. Sort it: measured, estimated or calculated.

3. Find the regulatory statement, if any. A valid statement names the country, the specific function, the intended use, the official source and the date. The FDA's general-wellness policy, dated January 2026, covers "low risk products that promote a healthy lifestyle", [1] so a product inside it has not been shown to work. An FDA 510(k) clearance means the device has the same intended use as a predicate device and is not identical to it. [2] The FDA also states that it has not authorized or approved any smartwatch or smart ring for measuring blood glucose on its own. [3] Clearance for one feature is not approval of every claim. Other countries have other systems; check the regulator for that country.

4. Look for validation by exact model. Look for the model and version, the people studied, the reference standard (for example ECG or lab testing), the conditions (rest, exercise, sleep) and the error numbers. Do not carry results from an older model or a different brand over. One umbrella review found about 11 percent of released wearables validated for at least one outcome, and that validation studies were just 3.5 percent of those needed for a full evaluation. [4] Results can also differ between devices for the same function, for example smartwatch atrial fibrillation detection varied by brand in a pooled analysis. [5]

5. Separate accuracy from benefit. A device can be accurate and still not improve your health. Ask for trials where people who used the device did better on a health outcome. For many consumer features, proper evidence not found is the honest answer.

6. Use manufacturer pages for specifications only. Sensor type, battery, sizes and stated ranges come from the maker. Accuracy, safety and benefit claims need independent studies.

7. Check the people it was tested on. Age, skin tone, body size and health conditions can change accuracy. See Wearable Validation in Different Populations.

8. Ask about your data. What is collected, who can see it, whether it can be shared or sold, how to delete it. Privacy protection differs by place and by who runs the app; see Health Data vs Consumer-App Data: Who Is Covered by Privacy Law?.

9. Decide what you will do with the reading. A reading should not diagnose you. If it worries you, or you feel unwell, talk with a clinician; in the US call 911 for severe symptoms. See Avoiding Self-Diagnosis From Device Readings.

What this site publishes and does not

• We publish explanations of what kinds of devices measure, with evidence read from studies and regulator pages.
• We describe a specific model only where current specifications and independent evidence both exist, and we say which model and version the evidence covers.
• Each page states the date the sources were read.
• We do not rank brands, give "best of" lists, state prices or accuracy scores we have not read from a source, or claim to have tested a device ourselves.
• We do not write or reproduce user reviews. Any reader review would be labeled as opinion, not evidence.
• When evidence is missing we say "Proper evidence not found" and what we checked.