Ambient Intelligence: Leveraging Computer Vision Technology to Improve OR Efficiency

Operating rooms are the site of some of hospitals' greatest success stories, but historically they haven't exactly been models of efficiency.

More than 70% of surgeries don't start on time, according to a 2016 study, a reflection of the less-than-scientific manual documentation common at hospitals. Start times, turnover intervals and prep milestones often depend on staff recollection or hurried chart entries. They're often wrong.

"If we're supposed to be in the operating room at 7:30 and we get in there at 7:36, it sometimes would get documented as 7:30," says Tony DeDominico, vice president of operations at Houston Methodist. "The recorded times often didn't match reality."

Those small discrepancies can add up, creating inefficiencies no one was pinning down. Delays upend hospital timetables, sometimes pushing surgeries way behind schedule and requiring staffers to work overtime, other times leaving valuable blocks of time unused.

Houston Methodist's new remedy: ambient intelligence. The technology uses embedded, unobtrusive technologies like sensors, cameras and AI to monitor and record OR activity from surgery prep to after-hours cleaning. Importantly, it provides predictive modeling, such as detecting when a case is expected to run longer than scheduled.

"There are a lot of different ways you can use the information," says Godfrey Ortiz, director of perioperative services at Houston Methodist. "We use it for efficiency of course, but it also generates real-time clinical insights and retrospective clinical quality reviews."

Houston Methodist launched the program, in partnership with San Francisco-based health care company Apella, in early 2023 as a pilot in the orthopedic and cardiovascular thoracic OR floors. It was rolled out systemwide in early 2024.

The early returns have been striking. Data from the pilot program shows:

• A 43% improvement in on-time first-case starts
• A 20% decrease in turnover times
• A 15% increase in surgical cases without added staff or space

The program uses four ceiling-mounted cameras in each OR and audio sensors in conjunction with patient data from the electronic health record. Using AI-driven computer vision, it watches for key events: when staff members arrive in the OR, when the room is ready, when anesthesia begins and ends.

From such signals, the system generates real-time alerts. A surgeon gets a text message when the patient is wheeled in, another when the drape goes up. After the procedure, the surgeon is notified that the patient has left for recovery — and when the next case has arrived.

The result is a smoother flow of communication. No more missed calls or delays in paging. The system simply keeps everyone in sync.

But one great benefit occurs after the fact. Leaders can review the data — not the surgery itself, but the surrounding workflow — to identify bottlenecks: Did the anesthesia setup slow things down? Was the patient prep time too long? Did room cleaning take longer than expected?

"It's just more of an overview," says DeDominico. "It's not about dissecting minute by minute or second by second, it's about looking at the overall flow of the room and figuring out where we can improve."

The program also improves safety. Video and audio let teams review potentially adverse events. In one case, it identified the cause of a "close call" in which a malfunctioning laser burned through the surgical drape; in another, recurring equipment failures were quickly diagnosed instead of requiring weeks of interview and guesswork.

But introducing cameras into the OR can raise eyebrows. Will patients be comfortable being recorded? Will surgeons feel scrutinized?

DeDominico acknowledges the early concerns. "We heard the questions — who has access, how long is it saved, what about privacy?"

Fortunately, safeguards are built into the system. Patients are made aware that multiple technologies are used throughout the hospital to provide safe and enhanced care and must consent upon arrival at the facility. Footage is automatically deleted after a short window, and review of recordings is governed by strict rules. Day-to-day users only interact with real-time insights; they don't comb through recordings.

"It's a tool, not surveillance," DeDominico stresses. "We don't save the recordings. We just use it to improve efficiency and patient care."

The potential benefits may not be limited to the OR. Possible future steps for ambient intelligence include expanding it into more inpatient hospital rooms, where integrating it with medical-grade wearable devices and smart sensors can lead to comprehensive remote patient monitoring for patient falls, distress or other safety-related events.

For years, hospitals counted on the EHR for its OR data. But the EHR is meant for record-keeping and billing, not management and analysis. Given ORs are likely the most resource-intensive spaces in a hospital, where every minute of delay or inefficiency affects patients, providers and the bottom line, innovation is much needed.

"AI is part of the future of health care," says Ortiz. "The OR is a complex environment, but we've proven this kind of ambient intelligence can work. The result is better efficiency, better safety and better care."

This article originally appeared in Innovation Spotlight, a LinkedIn newsletter on how Houston Methodist is leveraging technology to shape a smarter health system.