Pain relief for hospitals managing IoT performance

Nowhere is there more pain for IT staff than in the ever-morphing healthcare market where the Internet of Things (IoT) has been gaining attention and traction.

The concept of IoT involves the use of electronic devices that capture or monitor data and are connected via wireless to a private or public cloud, enabling them to automatically trigger certain events. 

In the healthcare context, a growing set of IoT devices have been introduced to patients and medical staff in various forms. Whether wireless bedside monitors, infusion pumps, or even voice/data-based clinician communication devices, the result is means better and more efficient patient care.

But these devices have more unpredictable “user experience” than their traditional wired counterparts and gaining visibility and insight into their performance has become a major issue for hospital IT staff. 

For example, if bedside monitors aren’t transmitting data related to patient vital statistics to viewers and central stations viewing that data, then the patient could be distress without clinicians knowing about it.  Similarly, if clinician communication devices are experiencing data loss or poor call quality, patient care can also be adversely affected.

Giving sight to the IoT blind

Recent technical advances in the areas of big data network analytics, machine learning and cloud computing are now poised to solve these problems, delivering much needed IT pain relief and unprecedented visibility into IoT network performance.

These innovations center on capturing, parsing and analyzing data emitted by healthcare IoT devices and automatically correlating this information across the full-stack of network data transactions experienced by any IoT client our group of IoT devices.

This approach stands in stark contrast to how traditional infrastructure management solutions have dealt with mission critical wireless IoT devices, such as bedside monitors.

Traditional infrastructure management has primarily used Wi-Fi infrastructure measurements to determine client behavior – often assuming that if the Wi-Fi infrastructure is performing well, then everything is operating properly. But this is not always the case as essential client device and application information is excluded from conventional infrastructure management models.

The key is having a big data analytics platform that collects application metrics around performance — Are the viewers getting good, complete heart rate, oxygen level, etc. data from the bedside monitors without any of it being dropped?  Are the clinician communication devices having good call quality?   Next, collecting the data isn’t good enough: the analytics system needs to automatically correlate this with client perspective data at lower layers, e.g. did the poor application performance occur when the client had poor signal-to-noise (SNR) ratio?  Or when the access point it was connected to had high levels of non-802.11 noise?

“Hospitals are spending too much time and money introducing new critical care IoT devices into the infrastructure to risk flying bind about their performance on the network,” said Brian Totten, Mobility Architect at Mission Healthcare System based in Ashland, North Carolina.

“If we don’t know how these devices are behaving on the network, we simply can’t use them with any sort of assurance. What’s desperately needed is the ability to capture and measure specific protocols from these devices and simultaneously analyze them against everything else going on in the network. This gives us a much more complete picture of IoT device performance that, until now, we just haven’t had,” he said.

In other words, the administration and monitoring of the wireless network today is completely separated from the administration IoT devices and applications. To the wireless network engineer, IoT devices have become a big blind spot, and vice versa for those responsible for device performance on the wireless network.

New device performance management platforms change all this using advanced data analytics to determine how, when and why waveform data, among other things, is getting lost by a bedside monitor while correlating that loss with the client’s current state on the wireless network.

By doing this, network staff can tell whether the data is getting lost because of poor RF conditions, the client just having roamed, or whether the issue is on the wired network or on the device itself. 

These systems leverage machine learning algorithms to identify trends and patterns and cloud resources to perform all the heavy computing chores in real time.

The data, already running over the network but just not accessible in any meaningful manner to IT staff, is then aggregated over a longer time period to establish which bedside monitors are the most problematic, what kind of issues are occurring over and over again, and what can to be done to proactively fix any individual or systemic device performance issues.

Armed with new IoT insights, healthcare IT teams immediately gain a view into device performance not previously possible. These include but are not limited to: the automatic root cause analysis of monitor connectivity issues, problematic monitors, the baseline behavior of bedside monitor performance and the amount of traffic monitors are sending and receiving.

Can you hear me now?

Bedside monitors aren’t the only mission critical wireless IoT devices used in hospitals. 

Today, physicians, nurses, assistants, and support staff often use a variety of different devices, such as pagers, multicast push-to-talk devices and even personal smart phones to communicate with each other. Managing the performance of every different device type on the network has become a massive challenge for IT staff.

Consequently, many healthcare institutions around the world are working to unify voice and data mobile communications among clinical staff using secure, purpose-built Wi-Fi-only voice and messaging solutions. Ensuring the best possible performance of these systems over the network is key to the successful integration, use and value of these new UC systems.

Many of these new systems, from vendors such as Ascom and others, now support the ability to output specific SYSLOG data from each Wi-Fi phone. This information can be used to quantify the quality and identify specific attributes of every call or communications made from those devices on the network. Access to and integration of this data into new network analytics platforms enables a similar client experience analysis and correlation with the wireless network for these devices alongside all the other devices analyzed.

This lets IT staff automatically determine, for instance, which clinician communication devices are experiencing poor voice call quality or frequent disconnections and if the problem was caused by the wireless network or some other client network transaction.

As more money and time is spent integrating IoT devices into enterprise access networks, it’s never been more important to justify these expenditures while ensuring the best possible user experience. The confluence of machine learning, data analytics and cloud competing is making this not only possible, but probable.