Proximie

Operating room intelligence on the platform runs on models: proprietary computer vision detects surgical events in the room and removes manual timestamp entry, and predictive models benchmark historical performance to forecast finish times, flag overruns and inform scheduling, with AI generated insights and recommended actions in the Daily Huddle view.

The Surgical Suite, the company's founding product, is telepresence and content. A remote expert, trainee or device representative joins a live procedure with video, audio and annotation, and recordings build a case library for training and review; that part of the platform sells on its own.

People make the operating decisions. A remote expert advises the surgeon in the room, and the Intelligence Suite feeds dashboards, daily huddles and notifications to perioperative teams and control centers, who decide on schedules, staffing and turnover. The system acts on its own in two places: it sends automated notifications along the care pathway and writes captured data back to the EHR without manual entry.

Whether staff can correct a detected event before it reaches the EHR is not stated.

The Intelligence Suite detects surgical events with proprietary computer vision on low resolution, frame based capture, and forecasts finish times with proprietary models that benchmark historical performance; edge machine learning runs on older devices. AWS supplies the cloud and AI infrastructure for the NHS evaluation, and Proximie works with NVIDIA on Project Rheo.

No model family, training data, accuracy figure or version practice is identified.

The computer vision and prediction models are Proximie's own, and the company owns and operates its stack on AWS, which also supplies cloud and AI infrastructure for the NHS evaluation. The privacy notice names AWS as host with region choices and the affiliates that receive data (Proximie Limited in the United Kingdom, Proximie SAL in Lebanon, Proximie Inc in the United States), and the data processing agreement gives ten days' notice of new subprocessors with a right to object.

Data also flows outward: de identified or anonymized data trains Proximie's algorithms, device companies receive structured procedure data to train their own, and Proximie works with NVIDIA on Project Rheo. Retention terms for those outward flows are not published.

Researchers have used the platform as the recording environment for more than 25 published clinical journal studies, and the company reports more than 500 facilities in 50 countries, 80,000 video views and more than 100 terabytes of video. Early Intelligence Suite deployments point to one additional surgical case per day, a figure tied to robotic surgery lists at Guy's and St Thomas', and a blog post describes a scheduling gap worth 19 percent of OR time. A year long NHS evaluation that began in September 2026 applies one framework across 104 operating rooms and procedure suites at eight hospitals, covering more than 20,000 procedures a year.

The throughput figures carry no baseline or period, and the NHS evaluation has no results yet.

Computer vision de identifies patient and staff data at the source, and in the NHS deployments data is de identified and anonymized before Proximie's models turn it into insight. The published business associate agreement (section 4.4) lets Proximie de identify protected health information under 45 CFR 164.514 and use it for any purpose, including training its algorithms, and the data processing agreement (clause 11.4) uses anonymized customer data to develop its algorithms. Device companies can draw structured data from real procedures through the Deploy AI use case.

Retention follows internal policy, contracts and customer instructions, with no period stated.

Proximie publishes its business associate agreement in full. It acts as business associate, or subcontractor business associate, to any customer that is a covered entity or business associate whenever it may access protected health information, and it must sign a subcontractor business associate agreement with any subcontractor or agent that handles that information. The agreement is governed by New York law, and a data processing agreement is published for GDPR customers.

Under section 4.4, de identified data falls outside the agreement and may be used for any purpose, including training Proximie's algorithms.

The platform is presented as complying with global standards such as HIPAA, GDPR and ISO, and a Security Addendum describing Proximie's security measures is available on request under a nondisclosure agreement. Proximie owns and operates its technology stack on AWS, with region selection at setup and single tenant systems available through its SDK.

No SOC 2 report, ISO 27001 certificate or NHS Data Security and Protection Toolkit status is named.

Telepresence, recording and operational analytics make up the platform, and none of its products is positioned as a medical device or carries a statement of intended purpose. Event detection and finish time forecasts inform scheduling, staffing and turnover rather than diagnosis or treatment, and the Surgical Suite connects people rather than interpreting the procedure. Data supplied through the Deploy AI use case helps device makers build products that follow their own regulatory paths.

Computer vision models in the Intelligence Suite are described as highly accurate, and data is obscured at the point of capture. The NHS evaluation that began in September 2026 applies a common framework across eight hospital sites, spanning operating rooms, endoscopy suites and catheterization labs, to measure operational effect at scale.

How event detection and duration prediction perform across sites, specialties, case mix and equipment is not reported.

Detected events, finish time forecasts, overrun flags, notifications and op note capture flow to teams and the EHR without manual entry, so the outputs reach schedulers and the record directly. The business associate and data processing agreements govern how data is handled, and surgical teams are measured on workflow and duration through the same analytics.

No accuracy figure, error rate, warranty or remediation commitment covers those outputs.

A live, two way data loop connects the Intelligence Suite to the hospital's EHR, with HL7 and FHIR support; it works with any EHR vendor and captures op notes automatically, giving perioperative teams a minute by minute view of cases in progress. The platform also connects to operating room setups and medical devices, and an EHR case study reports one additional case per OR per day.

No EHR vendor or live integration site is named.

Customers pick a hosting region on AWS at setup: US East in Ohio, Europe in Frankfurt, or Asia Pacific in Singapore or Sydney. Session video, audio, images, chats and telestrations stay in that region, while a Global Data Region in Frankfurt and Paris holds pseudonymized session titles, times and user identity for security and redundancy. Personal information may also go to the corporate group in the United Kingdom, Lebanon and the United States, and subprocessors may process data in any country under standard contractual clauses. Proximie owns and operates its stack and offers single tenant systems through its SDK.

Isolation between hospital customers on the standard offering is not described.

Hospitals and device companies buy through separate demo routes, with a dedicated route for Olympus. Capture hardware comes as Px Kit cameras and Vuzix smart glasses with a published warranty, an SDK serves device companies building their own products, and the platform is also offered through an AWS partnership.

No rate, unit or charging basis is published.

The platform covers operating rooms and procedure rooms, and the NHS evaluation extends the Intelligence Suite to endoscopy suites and catheterization labs. The company reports more than 500 facilities in 50 countries and more than 40 device company customers, and the Surgical Suite is available in about 20 percent of NHS hospitals. Because it captures the room rather than one procedure type, specialty coverage follows the hospital.

Training is a use case of its own, with remote observation, a recorded case library and asynchronous assessment for surgical teams and device representatives; its results are testimonials rather than measured learning outcomes.