AEGIS Protector — Case Study

Learns your normal

AEGIS Protector doesn’t compare you to an average stranger. It builds a personal baseline and keeps adapting to your lifestyle, so it knows what healthy looks like for you — per activity, hour by hour. Judging you against yourself is what keeps its alarms meaningful instead of noisy.

Knows what you’re doing

It recognises and blends your activities — rest, walking, exercise, sleep — and judges you against the normal for what you’re doing now. A high heart rate at the gym isn’t a false alarm; the same reading lying still is. Context is half of catching trouble early.

Anything-not-normal is flagged

Because it learns normal, it catches any departure — a novelty-detection approach (one-class SVM / SVR), not just a short pre-programmed list of diseases. New things it has never been shown still break through, which is exactly what an early-warning guardian needs to do.

Early cardiac warning — the K-wave

A dedicated channel watches for the J-wave / Osborn “K-wave” and ST changes — among the earliest electrical signs of an impending heart attack — before symptoms are felt. It flags a beeping K-wave watch alert and, on a real cardiac pattern, an ambulance-siren escalation.

An elderly person resting at home, a smartwatch glowing on the wrist and a phone glowing on the nightstand — The people who need watching most are often furthest from a machine — the wearable goes where they go.

Closer to you than your phone. In a world where more people feel alone, nothing stays nearer than the device on your body — worn through the day, through the night, through the moments that matter. You put your phone down; your AEGIS is always with you.

Human-in-command

It flags; a clinician decides. AEGIS Protector improves through governed retraining, never silent self-adaptation — the same discipline that runs across every AEGIS system. The wearer and the clinician stay in charge of what happens next.

Beyond the heart — personal safety

Because it’s on the body and always on, it can carry personal-safety functions too — including a silent duress mode for a robbery or a threat. One device, protecting more than one kind of danger.

Measured, not claimed

The models behind it are the interpretable Support-Vector family the AEGIS programme favours — SVM, Nu-SVM and SVR, extended with adaptive membership weighting — and every number that matters is measured subject-independently (the same person is never in both training and test), so the figures are honest rather than flattering.

Stated plainly, so it stays credible. This is a feasibility demonstration of the engine, built and validated on open research signals — not a certified medical device. Turning the K-wave early-warning into a device that calls an ambulance on a real cardiac event requires labelled clinical data and regulatory validation. The wearable flags; a human, and in time a certified pathway, decides.

Its clinical sibling. The same engine unlocks a hospital-desktop tier — AEGIS Clinical — where one clinician watches every patient at once on a population danger scale. Same device, different job.