Ocuvera
Ocuvera sells AI fall prevention for hospital inpatients. A wall mounted sensor with a three dimensional depth camera watches the bed and chair day or night, and computer vision models trained on more than 250,000 hours of video of real hospital patients recognize the movements that come before a patient gets up unassisted. Nurses get an alert on the Ocuvera phone app with a live stream of the room and can see, hear and speak to the patient from anywhere. A monitoring station can watch up to 80 rooms, Virtual Sitter Mode adds continuous video for patients who need it, and the Video Vault keeps footage of adverse events for review and staff education. Sensors move between docks in each room, and a portal reports usage and alert response times. Ocuvera sells it as a cheaper alternative to sitters, saying sites typically replace 25% to 50% of them.
A peer reviewed study in the Journal of Patient Safety, led by the University of Nebraska Medical Center with an Ocuvera coauthor, tested an early version in 15 small rural hospitals: across 13 with complete data, injurious falls fell from 2.52 to 0.55 per 1,000 admissions, with a median alert lead time of 28.5 seconds and 97.4% sensitivity. Case studies name Mercy Jefferson, Clark Regional Medical Center and CHI Health Lakeside. Ocuvera is based in Lincoln, Nebraska, and won a 2020 Microsoft Health Innovation Award in AI and machine learning.
Capability Axes
An AI Health Index grade measures what a buyer can verify from public sources on the date shown. It is not a rating of how good the product is. A vendor can build an excellent system and grade low on an axis because it publishes nothing an outsider can check. How grades read
Prediction is what Ocuvera sells: computer vision models read the depth camera's view of the bed and chair and estimate when a patient is about to get up unassisted, early enough for a nurse to reach the room. Ocuvera's own engineers describe the people tracking and bed exit likelihood models in a 2020 computer vision workshop paper, and the models were trained on more than 250,000 hours of hospital patient video.
Some of what a hospital buys works without the model: Virtual Sitter Mode's continuous live video, two way audio and the monitoring station make up a conventional virtual sitter service built around the predictive core.
The system only alerts, and a nurse decides what happens next. Alerts arrive on the Ocuvera app with a live stream of the room, so the nurse can see, hear and speak to the patient before going in; a monitoring station can watch many rooms and call patients, and nurses start, stop and pause monitoring at the sensor. The portal reports alert response times. What sets an alert off, how sensitivity is tuned per patient and what happens after a miss are not described.
A nurse's testimonial on the homepage says there are no false alarms, a claim with no figure behind it; the 2021 Journal of Patient Safety study of an early version measured a positive predictive value near 60%.
The approach and its training data are described. A three dimensional depth camera captures the room in light or dark, and customized computer vision algorithms trained on more than 250,000 hours of video from real hospital patients track people and estimate the likelihood of a bed exit.
Ocuvera's engineers have published two papers on the system, one modeling bed exit likelihood and a 2020 workshop paper covering people tracking, the video database and the move to Microsoft's Azure Kinect depth camera. Model versions, how often models are retrained and how a hospital learns that the model on its sensors changed are not described.
The models are Ocuvera's own, built and trained in house on its video library, so no outside model provider sits in the chain. What the chain looks like after the camera is not described: whether video and alerts pass through Ocuvera's cloud, a hospital server or a third party streaming service before reaching nurses' phones, which companies host the Video Vault footage, and who else can reach it. No subprocessor list is offered, and the privacy policy, last modified in December 2022, names no providers. The depth camera hardware comes from Microsoft, per Ocuvera's 2020 paper.
Evidence spans a peer reviewed study and named deployments. In the Journal of Patient Safety (2021), a University of Nebraska Medical Center team with an Ocuvera coauthor compared five baseline months with two alert months across 15 small rural hospitals: in nine using mobile units, unattended bed exits fell from 0.84 to 0.09 a day, and across 13 with complete data, falls fell from 8.83 to 5.53 and injurious falls from 2.52 to 0.55 per 1,000 admissions.
That study tested a prototype, and its authors call for research in larger hospitals. For the current system there are case studies from Mercy Jefferson, Clark Regional Medical Center and CHI Health Lakeside, 2022 statistical significance reports for medical surgical units, and unattributed results showing fall rate reductions of 35% to 64%.
Privacy rests on the camera: Ocuvera presents the three dimensional depth sensor as preserving patient privacy, and images on its site are labeled as deidentified. Beyond that, how patient video moves and how long it stays is not set out.
The Video Vault stores footage of adverse events for review and staff education with no retention period, Virtual Sitter Mode streams continuous video, and the models were trained on more than 250,000 hours of real hospital patient video with nothing saying whether customer footage still feeds training or under what consent. The privacy policy covers the website and services generally and never mentions patient data. How a missed alert is caught and reviewed, beyond the adverse event footage, is not covered.
No business associate status, HIPAA statement or agreement appears on Ocuvera's site. The only privacy document, modified in December 2022, speaks generally about people who visit and use Ocuvera services, covering cookies and contact data, and does not address patient video, alerts or Video Vault footage.
A system that streams and stores video of hospital patients handles protected health information, and the hospitals buying it are covered entities, so an agreement would be expected in contracting; none is described, and the subcontractors that would be bound by one are not named either.
No SOC 2 report, HITRUST certification, ISO/IEC 27001 certificate or penetration testing is claimed, and there is no trust center or security page. The only security language is the privacy policy's statement that Ocuvera uses the latest technologies and software to protect collected information.
For a system that streams live video of patient rooms to phones, offers two way audio into the room and stores adverse event footage, how streams are encrypted, how phones are managed and who can open the Video Vault are not described, so a hospital has nothing independent to rely on.
Ocuvera makes no device claim and states no regulatory position. The product predicts that a patient is about to leave the bed or chair and alerts a nurse, who watches the live stream and decides whether to go in, so the output is a prompt for staff rather than a clinical judgment acted on without a person.
The marketing says the system can predict falls before they happen and stop falls from the bed or chair, language that reaches toward fall prevention as a clinical outcome, and no statement explains how the product sits outside device regulation. It is positioned as monitoring support for nursing staff and a replacement for sitters.
How the models are tested, monitored or governed is not described. The 2021 study and the 2022 statistical reports measure overall effect, and the study found that a patient's age and fall risk were associated with the baseline rate of unattended bed exits, but no results compare detection across patient groups the camera may handle differently, such as patients using walkers, patients of different body sizes, confused or agitated patients, or crowded rooms with visitors. No review of model changes, monitoring of missed alerts in production, outside audit or policy on responsible AI appears.
Footage and reports give a hospital a way to question the system. The Video Vault keeps video of adverse events for review, and the portal reports system usage and alert response times, so a fall can be traced back to whether an alert fired and how fast staff responded. An error rate exists in the 2021 Journal of Patient Safety study, 97.4% sensitivity and a positive predictive value near 60% for an early version, but the site does not set it out as a commitment. No warranty, service level, credit or remedy tied to a missed alert or a fall appears, and no service terms are public.
No integration with an EHR, nurse call system, clinical communication platform or alarm middleware is named. Alerts run through Ocuvera's own app on phones Ocuvera supplies or the hospital provides, and through its own monitoring station, and usage and response reporting stays in Ocuvera's portal.
Fall risk scores, alert histories and monitoring sessions are not described as reaching the patient's record, and Ocuvera's monitoring has to be started and stopped at the sensor rather than from an order in the chart.
Hardware is described in detail: three pound sensors that click into wall docks in each room, carousels holding up to 12 sensors, and monitoring stations covering up to 80 rooms. Where the software runs is not. Ocuvera does not say whether video is processed on the sensor, on hospital servers or in a cloud, where alerts and live streams are routed, or where Video Vault footage and portal data rest, and no hosting provider, region or residency commitment is stated.
Ocuvera gives no price list, and a buyer reaches pricing through a demo request. Its site does frame the cost: a graphic puts an in person sitter at $18 an hour, a virtual sitter at $3 and Ocuvera at $2, a savings calculator estimates annual savings, Ocuvera says sites typically replace 25% to 50% of their sitters, and it promises savings the very next month.
Whether the $2 figure is a price Ocuvera charges per monitored hour or an estimate, and how sensors, docks, phones, monitoring stations and software are billed, purchased, leased or subscribed, are not stated.
Inpatient units are the setting, with patients at risk of falling from a bed or chair. The 2021 Journal of Patient Safety study validated an early version in 15 small rural hospitals, and Ocuvera's case studies and 2022 statistical reports cover medical surgical units at Mercy Jefferson, Clark Regional Medical Center and CHI Health Lakeside. Fall rates are tracked for unassisted falls from the bed or chair, the falls the system targets. Other units, such as intensive care, behavioral health or rehabilitation, and patients who fall while walking are outside what the evidence covers.
Alternatives to Ocuvera
Other products that do the same kind of work as Ocuvera, and what each one does.
care.ai makes virtual care and ambient monitoring software for hospital inpatient units.
VSTOne is VirtuSense's acute care product, a single sensor device per hospital room that predicts when a patient is about to leave the bed or chair unassisted and alerts staff before it happens.
Caregility, based in Wall, New Jersey, runs a virtual care platform for hospitals and other care settings.
Artisight sells a smart hospital platform built around one ceiling or wall mounted device that combines dual 4K cameras, a microphone array, radar, indoor positioning and an NVIDIA Jetson edge processor.
AvaSure makes virtual care technology for hospitals, and launched virtual sitting (TeleSitter) in 2009. A camera in the patient room feeds a hospital employed observer who can talk to the patient through two way audio and summon staff, and the company says one observer can cover up to 16 patients.
Vitalchat, based in Raleigh, North Carolina, sells a software led virtual care platform for hospitals.
Pricing
Vendor-published figures are labeled as such. Figures labeled “Estimated” are derived from third-party sources and have not been confirmed by the vendor.
| Entry Price | Pricing Basis | BAA Tier | Implementation | Source |
|---|---|---|---|---|
|
Not published.
|
Ocuvera prices after a demo request. Its site compares hourly costs, putting an in person sitter at $18 an hour, a virtual sitter at $3 and Ocuvera at $2, and says sites typically replace 25% to 50% of their sitters. How sensors, docks, phones, monitoring stations and software are billed is not stated. | Not published. | No installation or hardware fee is stated. Sensors click into wall mounted docks, and carousels hold up to 12 sensors. | Vendor Published |
A savings calculator on the site estimates annual savings from replacing sitters. Whether the $2 hourly figure is Ocuvera's charge per monitored hour is not stated.