Every conversation about healthcare security technology includes a discussion about AI, says Matthew Kjin, Segment Development Manager, Healthcare at Axis Communications.

Healthcare security

Artificial intelligence (AI) promises to give the healthcare industry – stretched dangerously thin – methods to cope with high demand without adding personnel.

These innovations can deliver efficiencies to every healthcare sector, including homecare, clinics, assisted living, inpatient and ambulatory services.

Due to the diverse nature of their operations, hospitals find that they can leverage the same AI models to address a range of challenges simply by initiating different workflows.

Depending on where AI is deployed and its intended purpose, a single analytics tool might be used to enhance healthcare security, improve patient support or increase operational efficiencies.

Remedying widespread healthcare security issues

Let’s start with a widespread healthcare security issue facing hospitals: property theft.

More specifically, let’s examine property theft in hospital parking lots.

Visitors and patients who visit hospitals may have certain sensitive items not allowed inside the facility, leaving them no choice but to leave these items sitting inside their parked cars for hours or days at a time.

What’s more, some people may leave valuable items in their car during an emergency or as they rush off to an appointment.

Therefore, would-be criminals know that hospital parking is the place to go if they are looking for crimes of opportunity.

They walk up and down the rows of vehicles, flipping handles and looking through car windows, oftentimes hitting the jackpot.

Two AI capabilities are proving particularly useful in reducing such incidents: direction detection and loitering detection.

Most lots and garages have surveillance cameras and license plate readers (LPR) that monitor the entries and exits.

People looking to steal will often enter the wrong way to skirt the LPR technology and get into the structure undetected.

Doing so may avoid the LPR reader, but AI-driven direction detection will alert security of the vehicle moving the “wrong way”, prompting them to send someone to investigate.

When live officers show up on the scene, they’re enough of a deterrent to turn away criminals and discourage future theft.

Loitering detection can also tip off healthcare security teams of suspicious behaviors linked to people looking for crimes of opportunity.

There are few good reasons why anyone wanders around a parking lot or hangs around outside a door for a prolonged period.

Maybe they’re just finishing a cigarette. Perhaps they’re lost and need help.

It’s also possible they’re waiting until the coast is clear to continue canvassing the lot.

AI doesn’t measure intent, but it alerts officers who can evaluate the situation and take appropriate steps. 

Solving a variety of cases with versatile tools

Hospitals can apply those same models, direction detection and loitering detection, to other use cases, including operational ones.

Direction detection may be incorporated into wayfinding solutions. Such technology helps visitors navigate confusing hospital layouts.

Some healthcare campuses offer dedicated smartphone mapping apps that make it much easier for those unfamiliar with the property to find a particular entrance, exit, wing, department or room number.

To augment those solutions, algorithms within cameras can detect people heading in the wrong direction – be it toward a restricted area or “in” through an exit-only.

This information can alert staff to intervene via strategically placed intercom devices. In some states, such solutions are becoming mandatory.

For example, Massachusetts recently passed Laura’s Law, which requires hospitals to deploy technologies like wayfinding, signage, two-way communication and monitoring to ensure patients can easily find the emergency room.

The legislation is named after Laura Levis, who died in 2016 outside a Boston-area hospital when she couldn’t locate the entrance to the emergency room.

Like direction detection, the loitering detection model has applications beyond looking for troublemakers in parking lots.

If a person hasn’t moved from outside the door to the neonatal ward for several minutes, they may be waiting to slip in the next time the door opens.

However, if a person hasn’t moved from the middle of the hallway in a surgical recovery ward, they may be having difficulty walking and need help.

Depending on the scenario, the same AI model can be programmed to alert a healthcare security officer in the former example and nursing staff in the latter.

Leveraging other technologies for safety, security and operations

People counting is another AI tool with many applications.

Overcrowding in the pharmacy may indicate a need to reallocate personnel resources to the register.

In a patient’s room, it could indicate unauthorized visitors in the building.

In a waiting area, it can indicate a need to provide additional seating.

In each case, AI can notify appropriate monitoring personnel or healthcare security teams so they can assess the situation and take the necessary action.

Another use for people counting is to secure drug storage areas, including rooms, cabinets and closets where controlled substances are kept.

Rules that require two people to be present whenever such drugs are removed reduces the incidence of theft by employees, as both would need to be in on the scheme.

In those same locations, AI can keep track of inventory and flag misappropriation of supplies.

Cameras running AI models can detect when stock is low or if an unusual quantity of an item is removed.

The same model can apply to drugs, face masks and bandages.

Utilizing AI to bolster the workforce allows employees to spend more time on patient and public-facing tasks.

LPR is an invaluable tool for protecting emergency departments and hospital staff.

While hospitals are obligated to treat everyone who shows up in the emergency room, they are not required to permit others to enter with them who may pose a security risk.

LPR can cross-check license plates with a watch list database, helping enforce healthcare security policies.

Similarly, workers who are concerned about domestic violence or other threats to their safety can ask healthcare security teams to use LPR to keep specific individuals off the property.

Radar is also used to secure emergency departments by monitoring traffic speed of cars approaching the adjacent drop-off and parking areas.

Frequently, a speeding car carries someone who has been involved in a violent crime, with the other people in the car wanting to avoid questions about how it happened.

The vehicle pulls up, an injured person is pushed out in front of the door and the car speeds off. 

Radar is capable of covering large areas.

When a car comes tearing in at undesirable speeds, radar alerts healthcare security teams immediately while providing live camera feed.

LPR captures the plate before the car speeds off, allowing law enforcement to jump-start an investigation.

Beyond the chaos of emergency rooms, medical staff also face physical threats from patients.

An astounding 73% of non-fatal workplace violence injuries occur within hospitals, according to an article in the Agency for Healthcare Research and Quality (AHRQ) PSNet Collection.

Many of those incidents occur without malicious intent; patients are confused, scared and may be heavily medicated, causing them to act out.

AI-equipped audio sensors can detect aggression without violating privacy rights that prohibit listening in or recording actual sounds.

Harmonic analytics measure the level and characteristics of the sounds in a room, looking for patterns that likely indicate brash or aggressive behavior.

Detection triggers an alert that dispatches appropriate backup personnel to de-escalate the situation.

Similarly, sensors in hallways, waiting rooms and other public areas can detect urgent, aggressive calls for help and alert healthcare security or medical staff, depending on the location and situation.

Creating a plan is key to a successful regimen

For hospitals to benefit from AI, they must first define specific applications and establish workflows to support its use.

Administrators must ask themselves, “What can AI detect or measure?” and then, “What can we program it to do with that information to make humans more effective?”

Then, they must carefully document each program to measure its efficacy.

This process requires metrics from before and after the implementation of AI.

How has its use impacted the incident rates of whatever it is detecting? How accurate has the detection been? Has it achieved the desired outcome?

After careful analysis, each program should be scored to support stakeholders’ efforts to promote adoption throughout their organization.

Higher scores indicate an investment delivers successful results and a higher return rate.

AI has the potential to help solve many problems endemic to hospitals, but the onus is on humans to build programs to leverage its power successfully.

Remember – AI alone doesn’t do anything. It requires planning, analysis and administrative work by thoughtful professionals before it can begin paying dividends.

About the author

Matthew Kjin, PSP, CPP, is the Healthcare Segment Development Manager at Axis Communications and a security and healthcare technologist with over 14 years of experience in healthcare services and life safety.

This article was originally published in the Special February Influencers Edition of Security Journal Americas. To read your FREE digital edition, click here.