← Back to News Device Safety

How we validate MCT-1 device and model safety

A health device is only worth wearing if you can trust what it tells you. Here's how we validate both the MCT-1 hardware and the models that interpret its signals.

Hardware and bench testing

Before a sensor ever touches skin, it's tested against reference instruments across temperature, motion, and wear conditions. We characterize accuracy, drift, and battery behavior so we know exactly how the device performs in the messy reality of daily life — not just on a lab bench.

Clinical correlation

Raw signals mean nothing until they're anchored to real physiology. We correlate the MCT-1's readings against established clinical measures so that a trend on your dashboard reflects something genuine happening in your body, not sensor noise.

Validating the algorithms

Every model is evaluated on held-out data before release and monitored for accuracy afterward. We track false positives and negatives carefully, because a wearable that cries wolf is as harmful as one that stays silent. Where the evidence is thin, the app tells you plainly.

Trust is earned in the details — the drift you correct for, the edge case you test, the alert you choose not to send.

Ongoing monitoring

Validation doesn't end at launch. We continuously watch device and model performance in the field and push improvements as we learn. Safety is a process, not a one-time checkbox.

Learn more about the MCT-1, read about responsible AI, or explore our research.