Method: how we log an instrument

What this site is about

Wearables and home measurement tools put numbers in front of people every day: heart rate, steps, sleep estimates, blood-oxygen readings, body temperature, indoor air quality, room humidity. Most of those numbers arrive without any indication of how accurate they are. We write about the science behind those gadgets: what a device actually senses, how that raw signal becomes the number on the screen, what independent researchers found when they checked it against a reference instrument, and where the readings stop being reliable.

We are not a product-testing laboratory, and we do not run our own accuracy trials. What we do is read: the manufacturer's specification and manual, the U.S. Food and Drug Administration's guidance and databases, and the peer-reviewed validation literature. Then we summarize what those sources say, with their limits left in.

The weekly issue

The site is organized as a notebook published in numbered issues. The stamp at the top of each page says which issue it was filed in, which week, and its entry number: instruments are numbered I-01, I-02 and so on, features F-01, F-02. When a page is materially revised, its "logged" or "last updated" date changes with it. Issue 01 (week 40 of 2026) is the launch issue, so every page published at launch carries that stamp.

What every instrument entry contains

How we read validation evidence

• Which model was tested. Validation studies often test an earlier generation of a device. We say plainly whether a study tested the exact model in the entry or a predecessor, and we do not carry a result across generations as if it were the same device.
• Against what reference. A heart-rate reading compared with an electrocardiogram chest strap is a different test from one compared with another wristband. We name the reference device and the setting (rest, treadmill, cycling, free living, sleep).
• How many people, and who. Small samples, narrow age ranges or a limited range of skin tones limit how far a result can be generalized. We report the sample size and say when a study population was narrow.
• Which error statistic. Mean absolute percentage error, mean bias with limits of agreement, correlation and, for pulse oximeters, the root-mean-square accuracy figure (A_rms) each answer a different question. We report the statistic the study used, not a converted one, and explain it in plain words.
• Who paid for it. Where a study discloses manufacturer funding or authorship, we note it.
• Regulatory status, checked at the source. A device is described as FDA-cleared only if we found its clearance in the FDA's own 510(k) database. Most wearables are sold as general wellness products, which the FDA does not review for accuracy; we say which category a product falls into.

What these gadgets are not

Consumer wellness gadgets are not diagnostic devices, and nothing on this site should be read as saying a gadget can tell you whether you have a medical condition. We describe devices only as estimating, recording or reporting a quantity. A reading that worries you is a reason to talk to a clinician, not to buy a second gadget. That caution sits on every entry on purpose.

How a product gets an entry

• An established brand with a live listing on Amazon.com. We open the listing ourselves and confirm the title, the brand, the exact model and variation, and that it is in stock.
• A manufacturer specification we can read. Sensors, battery, what the device reports, and the instructions for use.
• A price with a date on it. Every price on this site says when we saw it, for which variation and from which seller. A price is only true on that day.
• A validation record we can describe honestly, even when that record is thin. A product with little published evidence can still get an entry, as long as the entry says so.

How we make money

Science News Weekly earns money in exactly one way: the Amazon Associates program. As an Amazon Associate we earn from qualifying purchases. Every product button on this site goes to that product's Amazon listing, and if you buy through it we receive a small commission from Amazon at no extra cost to you.

There is no other affiliate program, no direct-seller link, no sponsored post, no paid placement, and no free product accepted in exchange for coverage. A brand cannot pay for an entry, and the commission rate plays no part in whether a product gets one or in what the entry says. The full version is in our Affiliate Disclosure.

What we do not do

• We do not run lab tests. We are a small editorial team without a laboratory. The accuracy strip summarizes other people's published measurements, and it says whose.
• We do not give medical advice. Nothing here is written for your particular situation, and nothing here can interpret your readings for you.
• We do not invent people. The byline on every page is Science News Weekly Editorial Team. There is no fictional doctor reviewing entries and no named expert, because we would rather you judge the sourcing than a face.
• We do not sell or ship anything. Every purchase happens on Amazon under Amazon's own terms. See Shipping and Returns.

Who writes this

Science News Weekly is operated by Aurora Group Platforms LLC, a Wyoming company. The pages are written and maintained by a small editorial team. We are not clinicians, engineers or metrologists, and we do not claim to be. What we bring is a process: open the listing, read the specification, find the validation studies and the regulatory record, date the price, write it plainly, and say when we could not confirm something.

Corrections

If something on this site is wrong, we want to fix it: a price that has moved, a study we described too generously, a result we attributed to the wrong model, a regulatory status that has changed, a link that has died. Write to us through the Contact page with the page address, the sentence or number you believe is wrong and, if you have one, a source we can open ourselves.

When we confirm an error we correct the page and update its date. If the correction changes what the entry says about a device, we add a short note on the page saying what changed.

How often pages are reviewed

Our working rule is to re-open every Amazon listing we link to at least once a quarter, confirming that the ASIN still resolves to the same product and variation and that it is in stock, and to re-check the price and the literature whenever we touch a page for any reason. Between reviews a listing or a firmware version can change without telling us, which is why the date on the page is the only date that counts.

Publisher