Can Health IT Platforms Support Success – Tech News| Tech News

The healthcare industry is frequently labeled as one of the slowest sectors to adopt new technologies, respond to consumer demands, and implement process changes that take advantage of the type of big data analytics that are revolutionizing retail, manufacturing, banking, and other segments of the national economy.

Most stakeholders would agree that the criticism isn’t entirely unwarranted.  A unique mix of regulatory constraints, ambitious reimbursement changes, traditionalist mentalities, and patchwork infrastructure development has made it exceedingly difficult for organizations of any size to turn on a dime.

Yet healthcare organizations should be given a great deal of credit for making significant strides over a relatively short time frame. 

In less than a decade, EHRs have become nearly universally used, if not universally beloved. 

In even less time than that, “big data analytics” has moved from a futuristic buzzword to a core component of every organization’s roadmap, even if they haven’t quite figured out how to integrate data successfully into the decision-making process.

And in what seems like just a few months, artificial intelligence and machine learning have exploded in popularity, moving from sci-fi to MRI at a blistering pace in some cutting-edge organizations.

This curious mix of rapid adoption and reluctance to abandon the status quo has led to some juddering starts and stops for many enterprises trying to grab a piece of this much-hyped “disruption” without actually disrupting the critical day-to-day operations of their organizations.

Promising pilots and passion projects are common in organizations that give their data champions a little leeway to pursue innovations, but rarely do these algorithms, risk scores, and predictive analytics tools spread beyond a single department or use case. 

Organizations may prefer to take a cheaper, quicker case-by-case approach instead of engaging with vendors who may require months of evaluation and implementation, tens of millions of dollars, and a complete infrastructure overhaul to start producing ROI.

“Becoming a technology company is pretty far from the core competencies of most healthcare organizations,” said Thomas Laur, CEO of SAP Health, to HealthITAnalytics.com. 

“But the barriers to entry with many of the traditional health IT vendors are very high, and the financial strain is simply not digestible for the ecosystem.”

The challenge has led many organizations to commit, reluctantly or otherwise, to fragmented, piecemeal development of one-off applications.  But that is akin to building their big data castle on sand. 

“While the application level seems very attractive because it delivery value quickly without a huge, rip-and-replace investment, it isn’t scalable,” he added.

Luckily, a middle ground is emerging.

Healthcare organizations need a firm foundation that allows for standardized development of applications that are both interoperable and intuitively designed, but they don’t always need to discard everything they have built over the past decade or longer.

“If organizations want to make those big investments worthwhile, they need to start thinking about developing technology platforms, not just building applications around specific, very narrow use cases,” said Laur.   

“Only a coordinated, standards-based approach to building backbone infrastructure will allow organizations to move towards data-driven decisions. I believe we are close to turning the corner on this.”

Health IT platforms, now available in all shapes, sizes, and specialties, offer organizations a way to enable coordinated application development without necessarily starting from scratch.

“Only a coordinated, standards-based approach to building backbone infrastructure will allow organizations to move towards data-driven decisions.”

When implemented well, the result is a standards-based environment in which to experiment with big data analytics and artificial intelligence to support value-based reimbursement, population health, clinical decision support, and countless other innovative strategies for improving quality while cutting costs.

Platforms don’t have to entirely replace existing, expensive infrastructure or well-worn workflows.  Instead, they allow organizations to take what they have and build higher than they ever could before.

While investigating the promises of a platform approach, healthcare organizations have to carefully weigh the benefits of committing to a specific ecosystem, develop the competencies to shift their perspective, and maintain a focus on interoperability so that they can share data outside of their own environments and ingest input appropriately.

What is a development platform?

In traditional software industry terms, a platform is any environment that allows for the hosting of applications or services. 

Platforms could take the form of a mobile or desktop operating system, a web-based browser or service and its application programming interfaces (APIs), a cloud computing framework, a set of programming specifications like Java or Flash, or even a piece of hardware on its own.

Within this broad definition, the variations are nearly endless. 

Facebook and Twitter are platforms, since they expose their APIs to third-party developers and allow for add-on apps to enhance their core services.  Amazon Alexa and Google Home have similar relationships with partners that link their own offerings into the ambient computing ecosystems to provide added value to users.

Marketplaces, like eBay or the Apple App Store, are also a type of platform, allowing for a centralized space in which parties communicate through mutually-beneficial standards.

Technology platforms, like Microsoft Azure, Amazon Web Services (AWS), or the Google Cloud Computing suite, allow developers to build unique applications on top of shared foundations that are portable, cost-effective, and scalable.

The attraction of platforms for healthcare is growing as value-based reimbursement initiatives demand that organizations extract more actionable insights from their data.

“In order to succeed as a participant in any type of value-based arrangement, you need to be able to look at everything about your providers, patients, facilities, and revenue.  You need to be able to drill down as far as possible, to the level of the individual, and get that 360 degree view that we’re all striving for,” said David Nace, MD, Chief Medical Officer of Innovaccer.

In the healthcare industry, electronic health records (EHRs) themselves are among the most ubiquitous platforms available, and are key to supporting the robust analytics that underpin value-based care. 

Bolt-on EHR modules are even a standard part of the health IT certification framework, offering the ability to customize, expand, and optimize a basic set of options for organizations with specific needs.

EHR vendors are also becoming hubs for third-party apps, creating algorithm bazaars like the Epic Systems App Orchard, the athenahealth Marketplace, and similar efforts from Cerner Corporation, Allscripts, and others.

For providers, the benefits of the platform approach are many.  Leveraging an existing application to run tailored enhancements requires little additional investment, allows users to stay within a familiar environment, and keeps workflows trim and streamlined.

Operating within a single environment is a big data boon, as well. 

If organizations can integrate and centralize their data assets and use one shared method for interacting with the data, users will be able to access more comprehensive insights with a lower risk of degrading the data’s integrity through repeated one-off transformations or multiple methods of input.

“To reach that point, you have to take a data-first approach, and ensure that your data is trustable, accurate, and as close to real-time as possible,” said Nace.

“That requires something to act as a platform so you can build out new applications, revise existing applications as needed, and really move organizations along their journey at the pace that makes the most sense to them.”

Many organizations are already heeding the call, according to a recent survey by Black Book. 

Having been burned in the past by data siloes and health information exchange barriers, the majority (70 percent) of healthcare systems in the market for new population health or revenue cycle tools are actively seeking an integrated platform that allows the seamless flow of data across the two interconnected areas.

In a separate survey by Spok released at the end of 2017, forty percent of hospital CIOs said they were planning to launch a healthcare analytics platform in 2018, with EHR integration, interoperability, and ease-of-use for clinicians among their primary goals.

The recent rush of cloud-based APIs, Platform-as-a-Service (PaaS), and Machine-Learning-as-a-Service (MLaaS) options from big players like AWS and Google are also indicative of a trend towards democratizing the ability to create innovative applications using the same set of plug-and-play tools as competitors and colleagues alike.

“It’s just like the fact that you have an iPhone or an Android in your pocket that you can use to install or delete apps as needed,” Nace explained.  “The apps are relatively disposable, and we’re always rethinking the functionality that we want our phones to deliver to us, but we couldn’t live very long without the phone itself, these days.”

“We keep that with us wherever we go, and we adjust its capabilities according to the situation we’re in.  I believe we’re going to be moving in that direction from an enterprise standpoint, as well.  The challenge will be to pick a platform that will enable you to move quickly and easily towards your goals.”

Taking a platform approach to tapping into the AI ecosystem

As artificial intelligence becomes more sophisticated and more desirable for organizations looking to improve productivity, safeguard patients, and reduce wasteful spending, the platform approach can bring an added benefit.

App stores and marketplaces that connect to mature environments can help to introduce organizations to algorithms and applications that might not otherwise make it onto their radar, says Abdul Hamid Halabi, Healthcare and Life Science Lead at NVIDIA.

“I talk to a lot of startups, and almost every single one of them struggles to get into the hospital environment on their own,” he said.  “The idea of going basically door-to-door with every hospital in the country is just not possible for a small company that’s starting out.”

“Smaller vendors need platforms that will allow them to scale, as well.  And in order to make sure these platforms are usable, they have to be based on standard architecture that can be easily shared and adopted by providers.”

The sheer number of available AI algorithms cropping up to address healthcare needs is both a blessing and a curse for developers and their target audiences, agreed Peter Durlach, Senior VP of Healthcare Strategy and New Business Development at Nuance.

“You have all of these people from every area of healthcare – and sometimes outside of healthcare – building these algorithms and tools, which is great,” he said.  “That is the goal of democratizing data.  But a lot of them have no way to get these offerings into production in the real world, and their ideas get lost in the shuffle.”

“And it poses problems for providers, too. You’re probably not going to listen to twelve different presentations and sign twelve contracts and business associate agreements (BAAs).  And you’re definitely not going to have much luck asking your IT department to take on a dozen separate implementation projects.”

Browsing through an app store integrated into an existing platform makes it easier for organizations to discover potentially game-changing applications and bring them into their environments without the time, expense, and hassle of dozens of separate procurement processes, Durlach said.

“You’re probably not going to listen to twelve different presentations and sign twelve contracts and business associate agreements.”

“Nuance is taking that approach with imaging analytics,” he said.  “We have a product called PowerScribe 360 that is used by 75 percent of radiologists in the US and an image sharing cloud-based platform called PowerShare, which is connected to about 5000 facilities.”

“Together, they basically create a two-sided network so that any publisher of an algorithm can single-click from any AI development environment into this cloud.  The consumers – in this case, the radiologists – can click to subscribe to these algorithms, just like you’d subscribe to a service on a smartphone app.”

Applications available through the Nuance Marketplace for Diagnostic Imaging, unveiled in late 2017, are validated in association with the American College of Radiology, he added.  The capability will be made generally available later in 2018.

“That’s the beauty of having an operating system, if you will, for a whole segment of your services,” said Durlach.  “It’s a very simple, streamlined way of connecting consumers and publishers without either of them having to worry about the infrastructure in between.”

“That’s generally something they are not prepared to do, and it’s usually a pretty significant barrier to getting these innovations into the workflow smoothly without a prohibitive cost.” 

Choosing a health IT platform to act as a broker between developers and end-users is going to be essential for organizations as they begin to rely more and more heavily on the advantages artificial intelligence can bring to their clinical care processes, stressed Halabi.

“Our bodies are very complex – there are so many different systems inside that all have multiple components and all work with each other,” he said.  “I can imagine that would mean that we need tens of thousands of algorithms to help make decisions around all of those points of interaction.  If you just look at a chest x-ray, for example, how many permutations of problems or abnormalities can you possibly see?”

“The systems we have now were not built to accommodate so many algorithms.  How do you integrate what artificial intelligence and deep learning can do into the systems we’re working with?  You need a data strategy, and you need a standard platform for deployment so that you can apply these algorithms to workflows in a meaningful way.”

NVIDIA is also working to bring enhanced applications to the imaging analytics world.  Alongside a partnership with Nuance on the PowerShare AI Marketplace, NVIDIA offers a hub for cutting-edge AI applications with its virtualized supercomputing platform, Clara. 

“We try to make it very, very simple,” said Halabi.  “You just tap into them, like you would tap into something on your phone or your PC.  You can say, ‘I want to run these five algorithms.  Here, send the data to them.’  It will run the algorithms against that data and bring you back the information.  That’s all you need.”

Simplicity is also the goal for Google, which has recently made a number of overtures to the healthcare industry – one of the last sectors with truly untapped big data potential.

In addition to sending Eric Schmidt, former Executive Chairman of Alphabet, Inc., to give a well-received keynote at the 2018 HIMSS Conference and Exhibition, Google used the annual gathering to unveil a slew of healthcare-focused announcements that build upon the company’s cloud-based offerings.

The Cloud Healthcare API combines powerful machine learning algorithms with Google’s widely used cloud capabilities, allowing healthcare organizations to work with their own data more easily.

“The goal of the API is to make it easy for health systems, providers, pharma, payers, to address their interoperability challenges,” said Joe Corkey, Head of Product for Health & Life Sciences at Google Cloud, at HIMSS18 in March.

“Google has been a long-time supporter of open standards, so the idea that we’re going to employ healthcare standards to make it easy for institutions to push their data to the cloud where it can be analyzed is a natural continuation for us.” 

“We’re also building in a lot of hooks to our cloud analytics and machine learning offerings.  The idea that once this data is in the cloud, in FHIR or DICOM or whatever standard you like, you will be able to run machine learning models against it.”

The strategy solves several of the industry’s most challenging problems around sharing data between disparate systems, said Aashima Gupta, Global Head of Healthcare Solutions at Google Cloud Platform.

“Interoperability is an imperative for healthcare,” she said.  “We don’t have to work very hard to convince people of that anymore.  But it is difficult to execute.  We can give organizations the tools to make interoperability work for them if they don’t have the ability to make them on their own.”

“With the cloud, now you can connect to apps.  The apps are interoperable.  Now it’s a platform.  Now you can do so much more than you could before, without putting all of the investment into scaling yourself.”

Once again, imaging analytics is a prime beneficiary of the platform approach, due in part to the fact that DICOM has been the indisputable and universal standard for imaging data for more than two decades.

“They can use the DICOM API to take the data from their PACS,” explained Gupta.  “We will do the transformation and put it on Big Query.  The hospital IT systems don’t have to do all that work – it’s done by our API.  Now the data is searchable and they can query it however they please.  Now they can get the insights they need.”

With great power comes great interoperability?

Nuance, NVIDIA, and Google – along with entities such as Amazon and Microsoft – are uniquely positioned to provide platform services because of their extensive reach, observed Durlach. 

“We can give organizations the tools to make interoperability work for them if they don’t have the ability to make them on their own.”

“There’s only a small number of commercial vendors that have the scale to engage in something like that,” he said.  “And the industry is very early in the process of creating a publishing universe, so it will be interesting to see who is going to win those battles as everyone tries to position themselves for success.”

“It’s not all about who can create the best algorithms.  It’s also about who can create the delivery pipeline for those tools, and who can create the best workflow experience, because that will come with a lot of opportunity to become the infrastructure organizations will rely on.”

The battle has already commenced – not just between established competitors like Google and Amazon, but also between electronic health record companies that are quickly metamorphosing into broader health IT ecosystem providers.

With the market share for foundational EHR systems almost fully absorbed by a handful of top companies, EHR vendors are now looking to bulk up their population health, clinical decision support, revenue cycle management, and business intelligence tools to attract new customers and retain the loyalty of existing patrons.

This race to become the clinical platform of choice for health systems and physicians has led many of the notable names in the industry to reposition themselves in the public eye, with Epic Systems being among the most forward to embrace the transformation from “electronic health record” to “comprehensive health record.”

“It will be interesting to see who is going to win those battles as everyone tries to position themselves for success.”

The term made a splash at HIMSS, where Epic Founder and CEO Judy Faulkner explained why a comprehensive clinical platform is necessary to enable proactive, holistic patient care.

“The comprehensive health record goes way beyond the 10 percent of care that physicians and hospitals directly contribute,” she said. “It incorporates genomics, environmental data, family history, socioeconomics – all of the factors providers will be responsible for as they stop getting paid for the 10 percent and start earning money based on keeping the patient whole.”

“When you bring a system like Epic into some of the places we need to be, such as a behavioral health program, you can allow those providers to add to the existing comprehensive record instead of creating new, fragmented records of care at different places.  Then you can share that information with providers at the patient’s care hub, whether that’s a hospital or a primary care provider.”

Epic’s extensive series of add-on tools help providers across the care continuum collect data and funnel it back into a single record for an individual, giving both clinicians and patients a seamless way to interact with the history of care.

“We’re treating everything as one system,” said Faulkner.  “It’s the first step towards making one unified record for every patient.  That’s what everyone has talked about, albeit in different ways.”

The thought is tantalizing, especially as the industry moves deeper into financial risk and outcomes-based payments, and clinically-focused health IT platforms could be an important vehicle for reaching that goal.

However, cautions John Supra, VP of Solutions & Services at South Carolina’s Care Coordination Institute, healthcare organizations have to be careful not to repeat some of the same mistakes that created hard-to-beat data siloes to begin with.

“It seems like for many of the big players, their solution is simply to say, ‘just bring everything into our systems, become our customers, and everything will be fine,’” he said.  “But putting all the data into one EHR doesn’t really meet the needs of a diverse network, especially if some of your members simply can’t afford to just switch over to one of the big guys.”

“If you’re just dividing the industry all over again into teams based around operating systems, you’re not really solving the problems of fragmentation.  You’re just shifting them around.” 

“And unless you truly embrace open standards and open architecture that allow for the free flow of data across vendor lines, I struggle to see how that interpretation of the ‘comprehensive health record’ meets the criteria that CMS Administrator Seema Verma has laid out in terms of open data, interoperability, and patient access.”

The concern is a valid one, especially since the industry’s ongoing conversation about interoperability shows no sign of flagging any time soon. 

Vendors deserve to be acknowledged for the strides they have made towards embracing standards, such as FHIR, and for actively bulking up the nation’s health information exchange (HIE) pipelines through initiatives like CommonWell and Carequality, and investing in growing health information exchange networks like Surescripts.

“If you’re just dividing the industry all over again…you’re not really solving the problems of fragmentation.  You’re just shifting them around.”

Regional and state-level HIE are also going strong as providers slowly embrace the data liberation movement.

But business incentives still tend to lean towards viewing patient data as an asset that should be closely guarded, and not everyone is convinced that sharing is a fundamental part of caring for patients.

“We need to break out of that attitude,” stated Supra.  “The only competition should be around the patient experience.  A patient shouldn’t have to choose where to seek care based on where their data is.  It should be the other way around.”

“So while we’re working on aggregating data and developing these operating systems on the clinical side, we absolutely have to keep our eye on interoperability, on standards, and on the ability to keep the patient in the middle by letting them control how and where their data is accessible.”

Durlach from Nuance agrees that both vendors and their customers have an obligation to discard the data hoarding mindset, especially if they wish to reap the rewards of value-based care.

“Data simply can’t stay siloed inside one hospital or one delivery network anymore,” he said.  “You will not succeed in this environment, which is moving so quickly on the reimbursement side, if you don’t leverage the data you’re sitting on.”

“You’re only going to maximize the value of your data by combining it with other data, and because of the way healthcare’s business incentives work, you need someone else to come in and aggregate that data and distribute it back.”

The key to connecting different operating systems is already known to the healthcare industry, says Google’s Aashima Gupta.

“FHIR can be the glue that connects your ecosystems together so that you can work with your partners effectively without reinventing the wheel.  It’s a powerful interoperability tool, and it is something that the vast majority of healthcare service providers, including Google, are working with very comfortably already.”

The internet-based standard is almost a platform in itself, and has quickly become extremely effective as the bridge between EHR giants like Epic, Cerner, Allscripts, and more.

As these platforms become larger and larger, FHIR will become even more critical for avoiding the formation of new data siloes and keeping the entire industry connected, no matter what OS allegiances a particular organization has embraced.

“FHIR can be the glue that connects your ecosystems together so that you can work with your partners effectively without reinventing the wheel.”

Interoperability must remain the focal point as organizations explore the promises that platforms have to offer, whether for big data analytics, artificial intelligence, or succeeding with value-based care.

Without seamless and trustworthy data exchange across platforms, organizations may find that their problem-solving skills are unable to keep up with the challenges of the industry.

“Your needs are always going to evolve,” said Halabi from NVIDIA.  “They’re always going to get more complicated, and there will always be a new issue to deal with.  The best way to prepare for that is to find a partner that can support you through your development in a very comprehensive, streamlined way that makes it simple for your providers to do their jobs.”

“That’s going to be a platform.  It might be multiple platforms, because healthcare providers have very complex needs, but you need to have access to environments that let you build as high as you need to.  Without that, it is going to be very difficult to get ahead of your competitors.”

A platform approach offers the right blend of simplicity and sophistication to enable organizations to take advantage of everything that big data and machine learning have to offer, said Gupta.

“Build it once,” she advised.  “Better yet, let someone else build it once.  Then you can take that standard and create something new that meets the needs you have when you have them.” 

“That’s what cloud can do, and that’s why platforms are so important.  You want one baseline that you can share with everyone in the same way, so that you can share your successes and move forward with confidence.”