AI is being applied in ever more exciting ways. With AI capabilities now embedded within video intercoms, network speakers and video surveillance cameras, the possible use cases are endless. We're now able to "teach" our intelligent network devices to understand what they are viewing, enabling them to analyze scenes and provide relevant data in real time.
This can benefit a wide variety of industries, one of which being healthcare. From patient care to operations, healthcare organizations can enhance a vast array of processes with AI. While this technology certainly will drive significant increases in efficiency and productivity, this comes from a place of enablement — not replacement.
Here are some key examples of ways in which Computer Vision based AI is already transforming healthcare.
Keeping track of supplies, patients and security
Hospitals are sprawling, open areas with an enormous number of patients, staff and visitors traversing the grounds at any given time. This means it can be challenging to monitor the entire property while giving every patient and staff member the proper level of attention. AI can help ease that burden by automatically detecting and alerting on pre-defined situations and events. The technology greatly aids situational awareness and can alert security personnel of aggressive and other types of concerning behavior in real time.
For instance, advancements in scene intelligence allow for very precise object tracking, which is incredibly useful for a variety of healthcare scenarios. Real-time object location tracking can track people and equipment throughout the property. This allows you to track patients who are at risk of wandering, including patients who may pose a risk of violence or who may become lost. If a patient is on the ground or unmoving for a certain period of time, a caregiver can be alerted immediately, ensuring optimal response time. Object location tracking can also be applied to loiterers and ultimately prevent criminal activity from taking place, such as infant abduction.
Healthcare units tend to have limited staff on hand, making it difficult to achieve optimal staff-to-patient ratios. AI is needed to help automate workflows and provide care teams with more time to tend to patient needs and other more pressing matters. For example, a common workflow for AI enables you to keep track of how many people are in a room (useful for detecting if the emergency room is overcrowded, or if there is a long line forming at the acceptance desk), how many cars are present or how much inventory is left. It can then notify when there's a low threshold of stock rather than requiring someone to go and check. Additionally, it can support teams with tracking an individual patient or resident.
Especially with workflows involving restrictive access, behavioral health patients, children getting lost and seniors leaving the premises, healthcare organizations can utilize object attributes such as clothing type for identification, while still respecting patient and resident privacy.
In areas such as a behavioral health or an ER where cameras are needed but consent for monitoring may not have been expressly given, the appropriate levels of privacy can be applied using AI. Fort Myers, Fla.-based Lee Health embedded AI-enabled cameras in the ER to blur out patients' features while still being able to see everything else in the room (e.g., number of people, placement, disposition, etc.).
This dynamic masking can also be used in infant hospitals and senior living accommodations to monitor if patients are standing or laying, or behaving in any other concerning way, while still easing concerns of invading privacy.
Safety as an Expectation
Many things can happen at once within a healthcare facility, and it can be near impossible for staff members to stay on top of every urgent scenario on their own. AI can therefore play a crucial role in promoting real-time awareness and promptly attending to urgent matters.
For example, object location functions like fall detection and "out of bed" detection can be used to alert when a patient is deemed at risk of a fall. The technology constructs a skeletal frame of the patient, ensuring that it can identify when a patient has fallen without the need to collect any personal identifiable information.
The same technology can be used to identify when a patient has gotten out of bed, as well as how long they have been out of bed — and whether they have fallen from a bed or onto the floor. This helps with quick responses when a patient at risk of falling is in danger, allowing staff to prevent falls or quickly get medical help if an injury occurs. Using this technology approach, a Wisconsin healthcare provider experienced an 80% reduction in falls during its pilot of a fall prevention program.
Meanwhile, sound detection AI can be used to automatically monitor for coughing and even measure the cadence and severity of coughs. The technology can also detect aggression and telemetry and physiological alarms, among other warning signs.
Proactive monitoring of healthcare environments
- Identifying a vehicle is as important as the ability to identify a person. License plate recognition can provide useful data on vehicles frequently lingering in sensitive areas. It can also help identify a vehicle involved in a crime, such as vandalism on hospital property; an assault or attempted assault on patients or staff; trespassing after hours; or drive-by crimes (such as picking up a patient against their will). This is all valuable evidence for both hospital security and law enforcement.
- Wrong-way detection uses object location to enable you to go beyond simply identifying a vehicle. Often, vehicle theft occurs after a vehicle enters the property in an unusual way. Perpetrators sometimes avoid normal entrances in the hopes of entering an area undetected. The ability to detect when a driver is behaving unusually or erratically can also help healthcare employees see when a medical event may be in progress.
- Weapon detection intelligence is also enabled through object location and can help healthcare facilities to automatically identify when a weapon, such as a firearm, may be present. If an AI model outside the building entrance identifies a possible weapon, it can alert security before the individual even enters the building. With proper procedures in place, the security team may be able to lock down the building and alert law enforcement in real time.
Bolstering the healthcare workforce
Today's AI capabilities can offer a great deal of assistance to healthcare employees. Whether it's to improve patient monitoring or drive more efficient staffing, adding embedded intelligence to your surveillance system can enable better patient outcomes and create safer, more effective facilities.
In deciding to leverage AI, healthcare organizations must also ensure it aligns with their governance, ethics and data protection policies.
Learn more about how Axis technology is supporting the healthcare sector here.