Half of older adults who lie on the floor over an hour do not survive the next six months. In a memory-care bathroom with nothing monitoring it, that hour runs until the 4 AM round. The bathroom is the highest-risk room and the one place a camera is not a usable option. Motion sensors can cover it, and off-the-shelf ones fire on the ceiling fan, the therapy dog, and every hard sit onto the toilet, until night staff mute the panel and nothing is watching at all. So we built Vigil, the intelligence layer between the sensor and the nurse call. It reads radar features, never a camera or an image, so the sensing can go where a camera cannot. We ran it on a seeded night shift over a synthetic radar feature stream: twelve labeled events, no real residents. Ten it suppressed, each with the number that decided it, like the resident who stopped at seat height instead of the floor. One it alerted on, the bathroom long-lie: the resident at floor level, breathing, no recovery, escalation armed from CNA to charge nurse to DON. And one real fall it would not call, because the room held two people and this version cannot tell them apart. That went to a staff check instead of a guess. If you run nights: which alarm on your panel did staff learn to ignore first? That one is the spec. #SeniorLiving #MemoryCare #FallDetection #AlarmFatigue #NurseCall