Improving User Experience with Healthcare Edge Networking, AI

July 12, 2018 - Healthcare poses unique challenges for IT infrastructure especially when it comes to solving challenges presented at the edge of the network. Understanding healthcare edge networking and how it’s impacting the industry is critical to future success, according to Extreme Networks Director of Product Management and Strategy Mike Leibovitz.

When clinicians are unable to access their digital tools, the consequences are much more serious than an average person who is unable to check their email. The healthcare industry deals with life and death scenarios which need to be heavily weighted when it comes to making decisions about the network. The network is a mission critical piece of infrastructure that supports the digital tools clinicians rely on to safely and accurately treat patients.

Many clinical and medical devices are connected both wired and wirelessly to a single network including tools used by clinicians, doctors, nurses, and biomedical researchers. Healthcare has also started transforming to more closely resemble retail and hospitality verticals. Patients and even visitors expect connectivity for data intensive apps like Netflix, YouTube, and FaceTime, explained Leibovitz.

While supporting personal patient and visitor applications may not seem critical, weak network connections can reflect poorly on an organization, which can in turn impact revenue.

“This challenge is tremendous because it’s about the end user experience of the individuals coming to your organization actually scoring your organization,” said Leibovitz. “HCAHPS scores include amenities like Wi-Fi or connectivity that come back to dollars and cents whether or not there's going to be reimbursement to that healthcare organization based on the score from that patient. Healthcare in the United States is a competitive scenario. A patient can choose to go to another healthcare provider down the street or across the road.”

READ MORE: Addressing Healthcare Network Connectivity Challenges

Providing clinicians with the best connections for their tools as well as satisfying patients prompts C-level executives to look at the network in more ways than just one. It’s not just about connected devices or security. End user experience is an important factor that has historically taken a back seat in the face of  issues that seem more pressing.

Putting all these factors onto one wireless network is a problem that needs to be solved. Taking advantage of edge networking can help distribute the network in a way that prevents many connectivity challenges.

“The edge of the network is something where end users, whether we're patients or clinicians, understand the connectivity,” explained Leibovitz. “We have a mobile device in our hand, an iPhone, a tablet, a laptop. We expect connectivity to work.”

“End users don't take the time to rationalize or really care if other devices around them are communicating,” he continued. “If you're one of 80,000 people in a stadium, you don't care if the phones are working for everybody else, you care about yourself. This attitude in a healthcare scenario is where things really get challenging. A clinician who is using a communication device is communicating on the same wireless spectrum as a patient lying in the bed watching Netflix.”

These things compounded on top of one another factor into network uptime and reliability. Organizations need to build a network that can support all the devices connecting to it that's also adaptable to how healthcare as a business is evolving and is secure enough to support PHI. These challenges summarize why healthcare is more challenging from an edge perspective than other industries because it has business and compliance demands that other industries don’t have to face to this degree.

READ MORE: The Future of Edge Healthcare Services and HIT Infrastructure

“Designing for uptime and having the reliability in both the hardware and the software that manages the network becomes very critical,” said Leibovitz. “You're designing networks to never go down. Even finding a window of time to work on the network in a healthcare organization is almost impossible.”

“We use the term ‘game day,’” he continued. “If we’re talking about providing Wi-Fi for an NFL stadium, game day is three or four hours on a Sunday. “In healthcare, it's 24 hours a day, seven days a week and it doesn't stop.”

“There are people that are living and, unfortunately, dying inside of these buildings,” he continued. “You don't have the luxury of calling a timeout. A Saturday night at midnight is likely the busiest time in the ER. IT is at home sleeping. Users and administrators expect that when the ER is full of people being admitted, the machines and the computers that are admitting people, monitoring them, and even the amenity of Wi-Fi for people that are waiting, all works perfectly.”

The more users, devices, and applications are added to the network, the more likely it is that the network will malfunction because there is too much traffic. There needs to be a level of automation and intelligence to the network to correct traffic behavior when it starts to bottleneck.

Data transactions from medical devices, MRI machines, infusion pumps, and EHRs are happening all at the same time on the same network. This makes prioritizing traffic and deciding what gets through first and which is most urgent a challenge. Seeking out tools that can help automate these decisions can prevent human error and quickly sort through traffic. This is where automation and artificial intelligence come into play.

Healthcare organizations can have hundreds or even thousands of access points and each access point has a view of the airspace around it and understands the clients that are on it. Because Wi-Fi uses radio waves, there are other devices that operate in the same airspace that can cause interference such as microwaves.

Other devices operating in that spectrum can cause an access point to drop. Traditionally, each access point can do its best to optimize the airspace around it individually correct what’s happening in it’s radius.

“Access points have had the ability to change their operating parameters automatically one at a time,” Leibovitz observed. “If an access point chooses to change one of its operating parameters, did that change help? Did it make things better? If it made things better, for how many devices did it improve connectivity for? These are fundamental questions. If those changes were successful and improved the quality of experience for all of the users, maybe those changes should be applied to the rest of the network.”

“What machine learning and AI does is it puts a brain on top of not one access point, but all the access points in the network,” he continued. “Now it can start ingesting a lot of information of what's actually happening on the network in real-time.”

When machine learning is applied in this scenario, this information can be digested, and the network can begin to understand everything that’s happening. The machine can make real-time decisions to change parameters based on similar behavior across various access points.

“The real point of this technology is that human beings can operate large complex networks, but it requires manpower or womanpower in real-time and that’s cost to a business,” explained Leibovitz.

“The reality of healthcare is you don't want to leave your network to chance,” he continued. “You really do need people, and that's why healthcare organizations typically have a larger IT staff than many other organizations. To debug and do all of these maintenance tasks, it's time-intensive and it's human-intensive. This is where machines can be employed to do the job that you probably don't want your staff working on.”

Organizations that are in the process of upgrading their networks to be able to prioritize signals need to consider the project from a big picture perspective.

“Organizations need to consider the growth rate of devices that are coming to their network, the applications, the expectations, the assumptions, and requirement gathering that is put into a network is going to result in the output,” advised Leibovitz. “If you don't have the right assumptions and the right requirements going into your design and planning, the result of your network is likely to be insufficient.”

Leibovitz found that some organizations skip this step and end up failing. Some provider organizations try to remedy network issues by placing more access points and switches instead of digging deeper into what is really causing the problem. If this critical planning step is missed, Leibovitz advises that organizations do not proceed.

Organizations also need to understand the medical devices that are being used on the network.

“Clinician and patient devices are obvious,” Leibovitz observed. “There’s always going to be the next generation of iPhones and Androids, but what about medical devices? When are they going to be upgraded? Making sure those pieces all come together is critical so when you do buy new hardware infrastructure, the next generation of devices and the last generation are all supported.”

Understanding the types of devices and prioritizing signals is critical to building a network that is reliable and secure. Optimizing edge networking and adding a layer of AI can help organizations build a flexible and dynamic network that will support current and future medical devices.