As a technology analyst, I have been closely following the ever-changing world of collaboration, with a particular focus on how AI is transforming every aspect of how we communicate and get work done. Most of the attention in this space is on horizontal use cases, where UCaaS is deployed across organizations to support a wide range of collaboration scenarios.
All cloud providers can address these needs fairly well, but when it comes to vertical markets, the use cases present distinct challenges, and need to be addressed with more specific capabilities. This spotlight article is about healthcare, arguably the biggest and most complex vertical for these technologies to address. The opportunity, of course, is massive, and several companies are investing heavily to own it.
Furthermore, being healthcare, the stakes are much higher than in the consumer world, so the performance expectations for vendors are very demanding. Over the course of a single article, my analysis will have to be high level, but hopefully enough to illustrate both the challenges and opportunities facing this sector.
The Central Challenge – Clinician Burnout
Collaboration means different things in different verticals, and in healthcare, the main scenario is with patients, where clinicians must share information across various stages of care. Direct communication with the patient is just one mode of engagement, but the central repository of data about the patient’s history upon which treatment decisions are made is the EHR – electronic health record.
The EHR has emerged as the standard data management model in healthcare, and in keeping with how digital transformation is impacting every sector of our economy, this represents the transition from analog, physical forms of information – paper, medical imaging, etc. – to an electronic format that can be easily shared and analyzed by the entire care team.
While this seems like a tidy, seamless solution, not every form of information is captured digitally, and nor is it all captured in a consistent fashion. On top of that, EHR data exists across many silos, and there isn’t an efficient way to integrate all of this, especially in real time, which is what clinicians need, especially for acute care situations.
Aside from this being a sprawling patchwork of data that requires a lot of effort to manage – not to mention interpret for providing patient care – the onus falls on clinicians to somehow pull all the relevant data together without missing a beat, while an anxious patient is sitting across from them in an office or possibly a hospital bed. If this sounds like what contact center agents must deal with to provide great customer service in real time, you’d be correct, so the challenge has similar root causes.
When the topic of the Great Resignation comes up, healthcare is right up there, and these challenges are one of the factors leading to burnout among clinicians. Healthcare facilities have been under tremendous stress during the pandemic to manage the caseloads, and the last thing clinicians need is to have manual bottlenecks like this to impede their ability to collaborate effectively and provide proper patient care. These challenges are very real, and are tailor-made for AI-driven solutions.
Two AI Use Cases for Healthcare
These collaboration scenarios call for more than what off-the-shelf UCaaS can provide, especially when dealing with massive volumes of patient and clinical data, often across disparate sources. To properly address these needs, the full patient care journey must be considered, and that’s where the opportunities for collaboration solutions – especially those driven by AI – become bigger than you may realize. In this context, there are two very different healthcare AI use cases to consider.
Use case #1 – voice AI
In much the same way real-time transcription has made workplace meetings more productive, this form of speech recognition brings significant value-add in clinical settings. The most pertinent use case is the initial interaction between patient and clinician, when symptoms are being described and important details are being discussed. On a practical level, voice AI can provide a more complete and more accurate transcription than what clinicians can typically capture by taking notes.
Beyond that, however, not only can AI capture the conversation in digital form, but also automatically enter each item into the appropriate section or field of the EHR. Given the complexity of medical jargon, machine learning plays a key role here, improving the accuracy over time, making it easier for clinicians to remain engaged with the patient without having to wait for the speech engine to keep pace.
Not only does this automation save time and effort for the clinician, but it allows them to focus solely on the patient for deeper engagement. This addresses one burnout factor by removing the stress of having to capture all this information accurately, along with spending more quality time with patients. Now patients will feel more at ease with the care process, and clinicians can see more patients and make the overall caseload more manageable.
While voice AI use cases show great promise in the healthcare sector, there are a few issues holding back adoption. The investment in voice AI can be difficult to justify given that these new data streams only add incremental value for improving patient outcomes, not to mention having a short track in terms of transcription accuracy to earn the trust of clinicians.
Voice AI has potential to add greater value, but in the healthcare space, most of the data that clinicians rely on to make decisions is visual – not verbal – such as reviewing imaging or test results. It’s also fair to note that other voice-based solutions that pre-date AI are well-established, namely legacy transcription applications and medical scribes.
These options are adequate for most speech-to-text needs, but are cost-prohibitive to use on a large scale. To date, the use cases for AI-based speech-to-text have been limited (or poorly understood), and despite some inherent advantages – namely ease of scaling and the ability to perform analytics across large volumes of data – these legacy options represent adoption barriers for voice AI, at least for the time being.
There is also a generational shift to consider, in terms of younger clinicians who are tech-savvy and comfortable with digital media. They will have high expectations in terms of how they interact with today’s software applications, and voice AI is a good example of this. The older generation will tend to rely more on the tools they know and trust, so this would be another factor for why newer applications like voice AI have had limited uptake so far in healthcare.
Patients are also impacted by these new technologies, but in a different way. Every vertical market has its particular technology lexicon, and in healthcare, some use cases for voice AI go by the acronym of ACI – ambient clinical intelligence. While speech-to-text applications are used in clinical settings to chronicle conversations between patients and clinicians, with the emergence of mobile broadband and wearable tech, sensors can now capture speech wherever the patient is – namely in non-clinical settings.
This represents a different use case for voice AI, but one that can support more forms for home-based healthcare, and that’s another way for technology to relieve some of the onsite caseload pressure that contributes to burnout. While these benefits are compelling for healthcare providers, the “ambient” nature of ACI means that patients can be monitored anywhere at any time, and that could feel invasive to many people. This raises a variety of issues around privacy, ethics and informed consent that will create further adoption barriers until addressed in a transparent manner.
Use case #2 – intelligent clinical workflows
When thinking about the value of UCaaS technologies in healthcare, voice AI has a role to play for making communications easier. Important as this may be, collaboration is a much bigger value driver in healthcare, where improving clinical workflows is the focus. The burnout we hear so much about in this sector is caused not just by the pandemic-led rise in caseloads, but also about how inefficient the workflows are.
This brings us back to the EHR, and the difficulty clinicians have in accessing the right data at the right time. Aside from all the time spent manually searching for data, time must be spent analyzing it to ensure it’s as complete and accurate as possible, after which more analysis is needed for diagnosis and treatment.
Time spent behind screens doing all this means less time spent with patients, where none of the stakeholders end up being happy. For patients, this can become an impersonal experience, leading them to lose confidence in getting the care they need. Clinicians become frustrated with a poor EHR experience, along with the stress of trying to manage so much disparate information. Finally, the healthcare facility suffers via lower throughput, where fewer patients are being treated, and at a higher cost.
All of this points squarely to the need for better EHR workflow tools, especially those that are AI-driven. While modern medicine has benefited greatly from technology advances, a natural by-product has been an exponential rise in the data associated with those breakthroughs. Clinicians are overwhelmed by the volume of data, not just with isolated data sets, but also the associations across data sets to make an informed diagnosis, early detection, identify new treatments, etc.
This problem set is being addressed by a variety of companies, where clinicians can leverage analytics to pull data from a wide range of sources, even beyond those they normally use. Over time, some AI can learn more about each patient’s history as well as the needs of each clinician – especially specialists – and can then predict likely outcomes based on treatment. Since we’re dealing with human lives, these are big dots to connect, and it’s easy to see why these solutions have been developed.
Complex Market Dynamics
If markets behaved rationally, that would be the end of the story, and these clinical workflow bottlenecks would have been solved by now. Instead, healthcare is being transformed just like any other sector by digital transformation and the race to the cloud. Dominance in the cloud is a necessary condition for big tech players in any sector, and since healthcare has lagged other sectors in cloud migration, things are really heating up now. One reason for lagging is complexity, both of the healthcare system itself and the data that clinicians rely on for treating patients.
Only now have technologies in areas such as clinical workflows and voice AI matured to the point where the leading cloud players are investing heavily in healthcare. Cloud alone is not enough for these companies to be successful; they also need the applications, especially those that unlock the full value of patient data, which is primarily EHR-based.
Among these big tech players, Microsoft is leading the charge, and interestingly, they have bet heavily on voice AI as the application to build on. With Azure, they have a strong presence as a cloud provider, but they need other pieces to be a strong healthcare cloud provider, and that came via their recent acquisition of Nuance. This was their second-largest ever acquisition to date, and that tells you how important voice AI is to Microsoft’s aspirations in this space – despite the fact that ACI is not widely-adopted yet. While the pain points around clinical workflow seem more pressing than voice AI issues, they have not invested as heavily in this space.
The three other cloud majors – Amazon, Google, and IBM – all have strong voice AI capabilities, but not to the extent of what Microsoft now has with Nuance. These companies are important to watch by virtue of their cloud offerings, but not for being clinical workflow leaders. To some extent, this reflects how hot voice AI has become across all sectors, and not just healthcare.
Voice AI – in its many forms, including conversational AI – has become a must-have capability, and for these companies it’s the most direct point of entry that builds on their cloud offerings. By comparison, the complexities around clinical workflow make this a higher risk application to build your market entry strategy around, and these vendors have already experienced some setbacks in this regard.
Even though I said there will only be room for a few big tech players in this space, there is another wildcard to consider beyond this familiar circle. Oracle is not a leading cloud provider in healthcare, and nor are they viewed as a player in speech recognition. That hasn’t deterred them, though, and they have chosen a different route to market.
Their recent acquisition of EHR leader Cerner is just as big and bold as Microsoft acquiring Nuance. While Oracle’s cloud capabilities can be very competitive in this sector, they need healthcare-specific capabilities such as voice AI or clinical workflow to compete directly with these other cloud majors. Instead of going with these applications, however, they chose an entirely different approach, and that’s what makes them disruptive here.
The EHR space has two dominant players – Cerner and Epic – and acquiring one of them instantly makes Oracle a big tech healthcare player. This move is really an end-run around the other big tech cloud players who have focused on the applications to enter this market. EHR is the central data source for patient records, and with Cerner in the fold, Oracle will have more control over leveraging that data from the cloud. Time will tell, however, if Cerner’s EHR data will remain accessible to other cloud providers – or whether this makes it easier for Oracle acquire applications capabilities like voice AI or clinical workflow to get on equal footing with these other players.
Conclusion – Need to Look Beyond Big Tech for Solutions
In terms of a takeaway for readers, these recent moves may give the impression that only big tech players have a role to play in better leveraging healthcare data to make life easier for clinicians, and better for patients. One issue is that Microsoft’s focus on Nuance may create lofty expectations for ACI and voice AI, and as with all things AI, the hype exceeds reality. There’s still a silver bullet halo around AI, but ACI—and voice AI in general—still has a way to go before it can be a big difference-maker in addressing clinician pain points.
The net impact is that many healthcare providers have taken a wait-and-see approach with voice AI, putting off investments in technology that is badly needed today. Another fallout from this mindset about voice AI is the shadow it casts over the other use case—clinical workflows. Big tech gets most of the attention in the healthcare space, and since they haven’t had much success with AI-driven clinical workflows, it would be easy to conclude there’s nothing out there for healthcare providers.
This of course, is one of the big dangers that comes with big tech; namely for smaller innovators to get traction. The problems around clinical workflows are big enough to attract plenty of players, and the healthcare market is ready to use these solutions now – if only they knew about them.
Since most of these players are independent, and not in the fold of big tech, it will take some effort for them to get on the radar of healthcare decision-makers. They are out there, though, and I have come across a few of these—and there no doubt are others. So, if you’re ready to consider offerings outside the big tech bubble, companies like PatientKeeper, Transformative Med and Wellsheet would be good places to start.