At Fellowship Square Mesa, an assisted living community outside Phoenix, staff used to log an average of 20 resident falls per month. Within six months of installing a radar-based ceiling device called Paul, that number dropped to about three. Across 11 Arizona senior communities now running the technology, the average monthly fall reduction stands at 66%, according to figures from the company behind it. It is one of the more striking numbers to emerge from the fast-growing field of AI-driven elder care technology — and even its inventor is careful not to oversell what’s actually driving the improvement.
A Smoke Detector That Watches How You Walk
Paul is made by Helpany, a company founded by Sandro Cilurzo, a former cybersecurity officer at a Swiss healthcare facility. The device mounts on the ceiling and resembles an ordinary smoke detector, but instead of watching for fire it uses radar to track how residents move — their gait, posture, stride length, and how steadily they transition from sitting to standing or bed to floor. There’s no camera and no microphone, a deliberate design choice meant to sidestep the privacy objections that have dogged camera-based monitoring systems in memory care settings.
The system flags early warning signs of fall risk before an incident happens, alerting staff when, for example, a resident gets out of bed with an unsteady gait at 2 a.m. Paul has been running since it launched in Switzerland in May 2019, and in that time has amassed more than 6 billion motion measurements from seniors, data Helpany uses to let the device learn each resident’s individual baseline while also applying patterns learned across its broader user base.
Arizona’s Rapid Rollout
Helpany began deploying in Arizona in January 2024 and now operates thousands of devices across the Phoenix metro area and Tucson, with expansion into three states. The results at individual facilities have been dramatic by any standard: Park Senior Villas reported a 67% reduction in falls, Fellowship Square Mesa 69%, and Westminster Village Scottsdale 72%. Bethesda Gardens Assisted Living and Memory Care in Phoenix recorded a 62% drop within just the first 90 days.
The Inventor’s Own Caveat
Cilurzo, to his credit, doesn’t claim the sensor alone deserves the credit. He points to the Hawthorne effect — the well-documented tendency for people to change their behavior, and perform better, simply because they know they’re being observed. Staff who know a monitoring system is tracking resident movement may simply respond faster or check in more often, independent of any AI-generated alert. Cilurzo argues the scale of the improvement across so many communities is too large to be explained by observation effects alone, framing Paul as one piece of a “holistic approach” to fall prevention rather than a silver bullet. Notably, the company itself acknowledges these results have not undergone independent, peer-reviewed validation.
Why the Stakes Are So High
The financial and human cost behind these numbers is enormous. Falls remain the leading cause of injury death among Americans 65 and older, and the projected annual U.S. medical cost of older-adult falls is expected to reach roughly $80 billion by 2030, up from about $50 billion in 2020. A single fall resulting in a hip fracture and surgery routinely costs $30,000 to $50,000 in direct medical expense. About 60% of nursing home residents fall at least once a year, and roughly half of those fall multiple times — numbers that explain why facility operators are moving quickly to adopt detection technology even before its effectiveness is independently confirmed.
Regulators Are Watching the Response Clock, Too
Part of the urgency comes from regulatory pressure that has nothing to do with the sensors themselves. Between 2023 and 2026, the Centers for Medicare and Medicaid Services tightened how fall events feed into the public-facing Five-Star Quality Rating system, and the Joint Commission has reaffirmed falls as a top-three sentinel event category demanding documented reduction programs. Detection systems like Paul are explicitly marketed as reducing the gap between when a fall happens and when staff respond — industry data suggests that gap can otherwise stretch beyond two hours during periods without routine rounds, versus under 60 seconds with an active alert system.
What This Doesn’t Solve
None of this addresses the underlying constraint driving much of the fall crisis in senior care: staffing. Detection technology speeds up how quickly a caregiver learns something has gone wrong; it does not add a single additional aide to a facility’s payroll or loosen the staffing ratios required by state regulators. Executives at facilities using Paul and similar systems describe the technology as augmentation, not substitution — a way to make already-stretched staff more effective, not a replacement for the people actually walking the halls.
What’s Next
Helpany’s expansion beyond Arizona into two additional states suggests operators are betting the technology will keep paying off even without independent validation studies. Whether that bet holds up under academic scrutiny — and whether the striking percentage drops persist as the novelty of a new monitoring system wears off — will likely determine if ceiling-mounted radar sensors become as standard in senior living as smoke detectors themselves, or fade as one more promising pilot that never quite proved its case.
