Overflowing emergency rooms. Scarcity of hospital beds. Equipment shortages. Insufficient staffing. In the COVID-19 era, hospitals are facing extreme logistical, clinical and budgetary stresses, so they must find ways to streamline their operations as much as possible, without compromising the quality of healthcare or customer service. This is no small challenge, because they serve hundreds to thousands of patients and visitors each day –in addition to physicians, nurses and staff. Hospital campuses are complex facilities with a variety of buildings and services that have to be managed, including parking lots, gift shops, restrooms, cafeterias, storage rooms, and, of course, diagnostic and treatment facilities.
To operate smoothly, safely, and efficiently, many hospitals have invested significant time and money into security staff and technology. For example, the vast majority of hospitals have implemented video surveillance systems to monitor the premises and to review video footage of incidents. Unfortunately, it’s virtually impossible to monitor every CCTV camera in real-time, and it’s difficult to manually review video evidence after an incident. Consequently, security staff may not see situations as they are evolving, and most video footage is never reviewed.
The solution to that problem is video content analytics, which uses Artificial Intelligence and Deep Learning to detect, identify, extract, and catalog all the objects that appear in video footage, based on classes and attributes such as gender, appearance similarity, color, size, and direction of movement. Such systems can also be used to investigate behaviors, such as interactions between objects, dwell times, and navigation paths. (As discussed in my recent blog post), intelligent video surveillance is often used in healthcare settings for increasing situational awareness via real-time alerts. Here I’ll address how video content analytics can also be used in these environments for reviewing past incidents to accelerate post-event security investigations and for collecting trend data to derive operational intelligence and drive intelligent strategizing and decision-making.
Video content analytics allows operators to search and review video footage across multiple cameras, and apply a variety of filters to review only the most relevant information. For example, in the case of a robbery at a hospital gift shop or if PPE supplies or medication were stolen from inventory, security officers can focus the evidence search on relevant areas, time periods and descriptions of the person of interest – as an example, they could filter video objects for a male person wearing green upper wear and black lower wear. Using this search and filter technology, security teams can review hours of footage in a matter of a few minutes, which dramatically accelerates post-incident investigation. Hospitals can also use the software to investigate and collect evidence for malpractice claims; quickly pinpointing the relevant footage based on the involved staff and claimants using search criteria or even face recognition.
Beyond video evidence, video content analytics can provide valuable data for driving more than investigations: Video surveillance captures critical insight about healthcare and hospital campus activity that, without video analytics, is not fully leveraged. By aggregating video data over time, intelligent video surveillance can visualize data and deliver reports for optimizing business operations, campus traffic flows and compliance with public health mandates. Users can customize heatmaps and histograms based on the specific datapoints needed, whether to illustrate when and where peak vehicle or pedestrian traffic times occur or uncover bottleneck hotspots, as well as which entrances are over- or under-utilized. Equipped with this information, hospital managers can make more informed decisions about staffing, signage, or construction and strategic planning.
Using people-counting analysis over a period of time, they may be able to forecast how many people to expect in a waiting room or emergency room during certain weeks or months, based on historical data. And, they may be able to identify the peak usage periods for certain areas of the hospital, such as restrooms, by counting the number of people who enter a restroom door over durations of time, so they can schedule restroom maintenance based on average traffic flows and expected increases in visitors.
Ambulance bay traffic over time can be analyzed, as well as parking lot usage, so security managers know quantitatively how many ambulances, cars, or people use particular lots during certain times of the day, week, or month. Gift shops can track how many shoppers enter the store, and compare that to the number of purchases. They can also see heatmaps of foot traffic in their stores, so they know which display kiosks or aisles receive the most traffic.
The COVID-19 pandemic is an unprecedented and daunting challenge for all organizations that are tasked with ensuring workplace safety and public health. Because they treat patients who are sick with severe acute respiratory syndrome coronavirus 2 (SARS-CoV-2), it naturally follows that healthcare settings are more likely to have high rates of Coronavirus in their environments, which means that their employees, non-COVID-19-infected patients, and visitors are at a much greater risk of exposure. Therefore, in healthcare facilities it is even more important to be able to detect whether or not individuals are wearing masks and accelerate contact tracing for infected individuals; video analytic-driven proximity identification is crucial to those efforts.
Because it can be used for forensic video data review, video analytics such as face recognition, appearance similarity and proximity identification can be leveraged to identify and track the activity of diagnosed individuals to identify other patients and employees with whom he or she has been in contact. By using video intelligence software for contact tracing, hospital managers can protect the anonymity of the diagnosed person , while also enabling them to quickly notify people who have been exposed and direct them to self-quarantine, as recommended, to protect others from possible infection. The technology can also rule out those who have not been in close proximity to a diagnosed person, so they can be more at ease and will not have to self-quarantine.
With mask detection and proximity identification capabilities – for measuring and quantifying the distances between individuals throughout time – non-compliance with mask wearing and physical social distancing mandates can be easily identified for making actionable decisions and determining solutions to better protect staff, visitors and patients. For instance, operators can analyze video to explore which times of day have more proximity and physical distancing violations to uncover business intelligence about traffic and crowding and determine how it can be resolved.
Hospitals already have very high standards for safety, service, and cleanliness, yet in these trying times they need technologies to help them prevent or solve common problems more quickly and effectively. Investing in video content analytics is one sure way to increase overall security, ensure patient and personnel safety, as well as optimize operations.
Signup to receive a monthly blog digest.