Predictive fall prevention
Haven uses passive in-home sensing and wearable intelligence to notice subtle changes in how your loved one moves, weeks before a fall occurs. No pendants. No buttons to press.
“Where home feels safe again.”
Why HavenReactive versus predictive
By the time a fall happens, the window to prevent it has already closed. Existing devices work the same way: wait for someone to hit the ground, then raise an alarm. Medicare pays for most of what follows, which is why prevention, not response, is where the value sits.
| Figure | What it measures | Reference |
|---|---|---|
| 14M | Adults 65 and older who fall each year | Economic burden |
| 1 in 4 | Older adults who fall in a given year | Warning signs |
| $80B | Annual healthcare cost of non-fatal falls | Economic burden |
| 41k+ | Annual deaths from falls, adults 65 and older | Predictability of falls |
| +41% | Rise in fall-death rates since 2012 | Predictability of falls |
| $18,658 | Average cost of a fall-related emergency visit | Economic burden |
| 67% | Share of those costs paid by Medicare | Economic burden |
The familiesWho Haven is built for
Haven is built for every kind of aging-in-place family, across generations, cultures, and homes. Choose a family to read what their supporter told us.
“Mamá won't admit when her hip hurts. Last month Haven flagged her walking speed dropping for four days straight. I called her that night and we got her to PT before the next flare.”
“Dad is proud. He wouldn't wear a pendant if his life depended on it. Haven is invisible, which is the only reason it works. He doesn't know it's there, and I sleep better knowing it is.”
“My parents live with us, but I travel for work. Beacon gives me one number every morning with a short sentence of context. Exactly what a loving child wants to know, and nothing more.”
“After Grandma's fall two years ago we almost moved her into assisted living. Haven gave us the confidence that she's steady enough to stay, and a warning system if that ever changes.”
The scienceThe two-to-six-week window
Gait speed is now recognized in geriatric medicine as the sixth vital sign. Walking speed below 0.8 m/s predicts not only falls but also cognitive decline, hospitalization, and mortality.
Modern machine-learning models analyzing temporal gait features from a single walking cycle reach 93.6% classification accuracy for fall risk in older populations. Haven's approach synthesizes environmental sensors, wearable accelerometers, and an optional vision tier, building a fuller movement profile than any single-modality system.
Measurable gait deterioration, meaning reduced walking speed, shortened stride length, and increased step asymmetry, can be observed 2 to 6 weeks before a fall event.
Gait & Posture, 2023
These changes are usually imperceptible to family members, and to the person experiencing them. They are not imperceptible to sensors running continuously, twenty-four hours a day.
The data has always been there. What has been missing is a consumer platform that gathers it continuously and acts on it.
Research librarySeven papers, written for families
TechnologySense · process · inform
Hardware you place, software you carry, and a boundary nothing crosses: raw sensor data is processed inside the home and stays there. Only derived insight ever travels.
Steadfast™
Plug-in sensor
Passive mmWave room sensing. No wearables, no cameras.
Stride™
Wearable overlay
Works with the Apple Watch, Fitbit, or Garmin they already own.
Vigil™
Vision · optional
Privacy-tiered computer vision for real-time gait analysis.
Anchor™
Edge compute hub
Every signal processed locally. Only derived insight travels outward.
Raw data never leaves homeBeacon™
Family dashboard
Risk score, gait trend, and context, on the phone of whoever cares.
Beacon™
Family dashboard
What families open every morning. One risk score, alerts with context, and a weekly narrative written in plain English, never a clinical chart.
Anchor™
Edge compute hub · Matter-compatible
The nervous system of the home. Every signal is processed on the device; only derived insight leaves the house.
Steadfast™
Passive in-home sensor
24 GHz mmWave radar sensing for presence, movement, and transitions. No cameras, no microphones. Plug it in and it learns the rhythm of the home.
Stride™
Wearable overlay
Turns the Apple Watch, Fitbit, or Garmin they already wear into a continuous gait-baseline engine. No new hardware to buy.
Vigil™
Vision · optional tier
Computer-vision pose estimation for higher-risk situations. Privacy-tiered, opt-in, and controlled entirely by the family.
How prediction worksBaseline to warning
Haven does not wait for events. It learns one person's rhythm, then tells you when that rhythm changes.
Weeks 1–4
Stride observes gait, cadence, and sit-to-stand time. Steadfast learns sleep and movement patterns. Vigil, if enabled, reads posture and joint angles.
Ongoing
Plug in the hardware and forget it exists. Haven compares each day against a rolling 28-day baseline built from one person, not a population.
Early warning
Stride shortens. Cadence slows. Night-time restlessness or bathroom visits climb. Small shifts, days in a row. Haven notices the pattern.
Intervention
Beacon tells you what changed, in plain language, while there is still time to act. A call, a doctor's visit, a rug removed.
For the familyWhat a supporter actually sees
Beacon gives you one daily risk score in language anyone can read, so you can decide whether anything needs doing today.
No dashboards to interpret. No guessing from a weekly phone call. Just enough to know when to pick up the phone.
A Tuesday morning in Beacon“Mom has been more active this week, but restless for three of the last four nights. It might be worth a call.”
Peer-reviewed evidenceFindings, not marketing
Haven's architecture rests on published work in sensor-based activity recognition and gait-based fall prediction. These are findings from the literature, not measurements of Haven's shipping product.
| Finding | What was measured | Source |
|---|---|---|
| 98% | Fall-detection accuracy using mmWave radar with semi-supervised learning | mmFall · IEEE Trans. Automation |
| 96% | Gait step-time measurement accuracy mapping FMCW radar in real environments | PMC / Sensors · 2022 |
| 125 | Volunteers aged 6 to 63 tested, validating the mmWave gait ecosystem | RDGait · ACM UbiComp · 2024 |
| 93.6% | Classification accuracy from a single walking-cycle gait signature | Frontiers in AI · 2024 |
| >90% | Activity-recognition accuracy across eight distinct environments using Wi-Fi CSI | EHUNAM · Nature Scientific Data |
| 1.2TB | Public CSI and BFI dataset spanning 20 activities in complex non-line-of-sight scenarios | CSI-BFI-HAR · IEEE DataPort |
QuestionsWhat families ask first
Life Alert, Medical Guardian, and Apple Watch all register a fall after it happens. They are reactive. Haven is predictive: it notices subtle changes in gait, movement patterns, and daily routine to identify rising fall risk weeks before a fall occurs.
Haven also asks nothing of your loved one. No pendant to wear, no button to press, no device to charge or remember.
Haven's core system, meaning Steadfast sensors, the Stride wearable overlay, the Anchor edge hub, and the Beacon dashboard, uses no cameras whatsoever. Vigil is an optional computer-vision tier for higher-risk situations, and it is privacy-tiered and controlled entirely by the family.
All processing happens locally on the Anchor hub inside the home. Raw data never leaves the premises.
Peer-reviewed research shows AI models analyzing gait parameters reach 90–95% sensitivity in identifying people at elevated fall risk. A 2024 study in Frontiers in Artificial Intelligence reported 93.6% accuracy using temporal gait features from a single walking cycle.
Haven's multi-sensor approach across environmental, wearable, and optional vision is designed to exceed single-modality accuracy by building a fuller movement profile. These are literature findings, not measurements of a shipping Haven product.
Haven is built for adult children caring for aging parents who want to stay in their own homes. Your loved one is the person Haven learns; the daily reader of the dashboard is usually the family supporter, often a 45-to-65-year-old professional managing care from another city.
It is designed to give peace of mind without costing your parent their independence or their dignity.
Haven is in active development and launching in 2026. Waitlist members get first access to pilots, founding-member pricing, and direct access to our clinical team during the pilot phase.
The rate you join at is the rate you keep. Haven does not raise the price on an existing subscriber.
The consumer product is a wellness platform, not a diagnostic device. Vigil is being pursued on an FDA 510(k) pathway as a Class II device for higher-acuity and institutional settings, using the BioSensics LEGSys as predicate. A regulatory pathway has been identified; clearance is additive upside and is not a requirement for consumer launch.
Get in touch
Questions, partnerships, or press. We read everything. You will hear back from a founder within one business day.
Haven Home Wellness, Inc. · Atlanta, GA
info@havenhomewellness.ai
Legal
The full Terms of Service for Haven's products and website are available on request. Haven is in pre-launch development; platform terms will be finalized before public release.
For questions about terms, licensing, or commercial partnerships, write to info@havenhomewellness.ai.
Haven Home Wellness, Inc. · Delaware C-Corp · Atlanta, GA
