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.
- Reference comparison — validating against trusted clinical instruments.
- Real-world conditions — testing during activity, sleep, and flares, not just at rest.
- Population diversity — checking performance across different bodies and skin types.
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.