How Hospitals Are Adopting Artificial Intelligence to Safeguard Patients, Staff

For example, camera systems that integrate artificial intelligence software can help healthcare teams detect firearms and provide notifications when they spot weapons, Viner says.

“We're certainly looking at leveraging that sort of technology to identify when we have a threat,” he adds. Viner has 30 years of experience working in law enforcement.

AI systems can trigger alarms when a patient’s actions cause concern, says Mike Jude, research director for IDC’s video surveillance practice. “Recognizing violence can be as simple as flagging somebody who is moving inappropriately within an emergency room setting,” he says.

READ MORE: Why do physical security and cybersecurity work so well together?

HonorHealth Uses AI to Monitor Camera Footage

Before HonorHealth bolstered its physical security in 2018 by installing a video surveillance system, it relied on analog tapes.

Today, the health system has about 2,000 cameras. Every entrance, exit and common area has cameras, along with intermediate distribution frame closets, pharmacies and other sensitive areas around the hospital.

HonorHealth is now exploring AI assistance to sort through camera footage, according to Todd Larson, a consultant in workplace and public safety for the health system. “We need augmented help that can alert us on things so that we don’t have to spend time going through every one of these cameras or having to backtrack,” he says.

AI can spot irregular events, such as a visitor wearing a thick coat in 100-degree Arizona weather, Larson says. And while face recognition is a helpful tool, the ability to also recognize objects — such as an empty wheelchair in the middle of a hallway or an unattended wet floor— can also help prevent accidents.

“If that alert comes to our system, we can call security to put out a cone or put out a fan to dry that hallway,” Larson says. “Whatever that situation would be, we can remedy that fall before it happens.”

The next stage for HonorHealth’s video surveillance system would be to incorporate predictive analytics to better support security staff.

“We have these great camera systems, but there is so much more we could be doing with them,” Larson says. “With the right AI solution, they can help us by being proactive instead of reactive.”

Viner would like to set up a real-time operations center within the health system’s network operations center to monitor a large mass of cameras and deploy resources quickly in response to alerts. In the NOC, dispatchers and private branch exchange call-takers manage all the incoming calls for the network and dispatch from that location. Analytics and computer-vision technology may be added to camera systems to help hospital security spot threats.

“I think the general theme is to keep patients, staff and everybody involved safe,” Viner says. “The analytical part is to be able to identify patterns and respond to those patterns in a way that's useful. We're searching for what sort of software would be best suited for the surveillance systems we currently have in place.”

AI systems pick up ambient sounds, such as screaming in a waiting room. After the noise detector picks up the sound, it alerts the camera along with the hospital’s NOC. Dispatchers can then send security to respond to the disturbance, Larson says.

Sound, computer vision and face recognition are key components of an AI-powered video surveillance system. “Those three areas are critical to security and making sure that we have the ability to monitor properly,” Viner says.

RELATED: Physical security platforms support a growing number of use cases.

Yakima Memorial Uses Video Surveillance to Support Its Security Team

In the Pacific Northwest, MultiCare Yakima Memorial Hospital in central Washington decided to protect its multiple facilities by installing a Verkada video surveillance system.

ER foot traffic is the biggest threat to patient and staff safety, with 300 to 400 patients stepping into the department each day. Sometimes, patients become contentious with nurses during triage. The video system helps the Yakima Memorial security team monitor the environment and respond as needed.

Previous systems had poor image quality and could not distinguish a person from a pet, according to the health system. Another issue: The recording system would overwrite footage, making retrieval impossible.

To resolve these problems, Yakima Memorial turned to a Verkada hybrid cloud system incorporating advanced AI across seven physical locations. Additional sites, such as the North Star Lodge Cancer Care Center and Children’s Village, also have installed the Verkada cameras.

With computer vision features in the outdoor cameras, the health system can monitor areas just outside its facilities, seeing details on vehicles such as license plates.

When MultiCare Health System completed its purchase of Yakima Memorial earlier this year, there were discussions about using the hospital’s solution across the rest of MultiCare’s large organization, according to Bob Benoit, MultiCare’s CTO and vice president of digital transformation.

“In our hospital systems, crime and patient safety have been a growing concern. Our physical security team is doing an end-to-end assessment on ways to mature and advance our best practices to make the hospitals as safe as possible for both our patients and employees,” he says. “We think technology is going to have a big role in doing this with scale and cost-effectiveness.”

In the meantime, the health system deployed Verkada in the entryway and some operating rooms at MultiCare Capital Medical Center in Olympia, Wash., Benoit says.

“We’re in a learning phase where we’re using this modern system and out-of-the-box functionality for quick and simple wins, and then we’ll start turning on more-advanced functionality over time as we learn about the system and integrate it into how we work,” Benoit says.

Some video surveillance systems integrate with other applications such as badging systems, and the applications then “learn” from each other. Such large learning models bring detection, analysis and alerting capabilities that are faster than what a human can do, Benoit says.

For example, modern video technology lets health systems respond to incidents more quickly by spotting details and allowing for quick analysis, such as tracking a person’s location based on a specific article of clothing, according to Benoit.

“You can see how using generative AI to quickly identify risks can reduce the time it takes to respond and make better use of our security staff’s time,” he says. “We think that’s a big piece of the future.”