Features

Exploring Decentralized Architecture Networks in Healthcare

Experts from Sentara Healthcare outline decentralized architecture networks and data fluidity, sharing how these can be used to tear down healthcare data silos.

Source: Getty Images

- Efficient and secure health data interoperability is a top priority for providers, payers, and other healthcare stakeholders, but enabling interoperability remains a major challenge for many organizations across the country.

To support healthcare interoperability, various tools and concepts have been developed and promoted, such as health data standards for aggregation and normalization, including the Fast Healthcare Interoperability Resource (FHIR) and its accelerators, alongside compliance rules like the ONC interoperability rule.

Some of these resources can help healthcare organizations break down data silos, a critical factor in advancing interoperability, but other, less well-known tools also exist that stakeholders can leverage.

One of these is decentralized architecture networks, according to Dave Torgerson, vice president and chief analytics officer, and Tim Deutsch, vice president and chief data officer at Sentara Healthcare, who outlined what a decentralized architecture network is, how it can be used to enable data fluidity, and how that process can break down healthcare data silos in an interview with HealthITAnalytics.

DECENTRALIZED ARCHITECTURE NETWORKS AND DATA FLUIDITY

Before jumping into how decentralized architecture networks can help break down healthcare data silos, one needs a basic understanding of these networks and data fluidity, including how the two are related.

“Generally, most networks today are considered more of a centralized network where you have one master server that manages all the activity, and everything needs to work within that,” Deutsch stated. “But in a decentralized server, you have multiple servers that manage the information, and these individual servers can act as individual master servers by themselves, but they connect to each other. They can balance the load and distribute the work across the system.”

This ability of independent servers to act as master servers while still being connected creates a high degree of 'fault tolerance,' or resiliency within the network, Torgerson added.

“One node goes down, and the rest of the nodes can operate, in some cases, as if nothing's happened. So, you have the potential for fundamental [data] redundancy,” he explained.

Data redundancy, a term often discussed within the context of blockchain and cryptocurrencies, refers to the practice of keeping data in two or more places within a database. This practice can help ensure that no data corruption or loss occurs if implemented appropriately.

“Having replicable ‘sources of truth’ in different places so that you can maintain integrity, and you can't change one to deceive the others, are practical benefits of the nature of the [decentralized] architecture,” Torgerson continued.

The idea of data fluidity is another one of the major benefits that can come from the use of a decentralized architecture network. The concept describes the degree to which data are accessible and available.

“The basic idea here is that data is available universally in whatever form it's needed, in whatever place and time it's needed. So, whatever [data] I need, however I need to consume it, I'm able to do that. I'm not tied to necessarily a specific technology or a specific channel,” Torgerson explained. “We might think about it as — if I want to get data via an [application programming] interface, I can. If I want to get data via a database connection, I can. If I want to get data via an XML feed, I can… All that stuff is universally available to me, so I don't have to do too much management of my interfaces to get access to the data I need on the demand side.”

On the supply side, data are funneled into a ‘common environment,’ which makes them available in one place and minimizes the need for the user to accommodate different interfaces or match the data to a source before they can use them.

This can be particularly helpful depending on the use case or the data user, Deutsch stated.

“If you think about data and the maturity of data as a factory, like a manufacturing line, you take raw data, and you start building on it to make it more valuable to the consumers,” he explained. “And then you get to an endpoint where you might have a curated dataset where somebody can just pull information that's already been integrated and combined in a way that is good for [your] business. Well, different users, different use cases have different needs for different types of data.”

For example, Deutsch noted that a data scientist might have a particular use case where they need to look at already aggregated, mature data. From there, they may need to trace the development path of that data or look at the raw data. This allows them, as the user, to find answers to their data-related questions at any point along the data’s development.

The ability to find where the data originated or determine how the data were developed is crucial in an industry like healthcare, but conventional data storage and sharing approaches can hinder this process.

TEARING DOWN HEALTHCARE DATA SILOS

Healthcare data silos can present a significant challenge for healthcare organizations looking to efficiently analyze data or locate the ‘source of truth.’

“[Data silos] make it harder for anybody to find where the data is because you often find that the same information is stored in multiple silos, and you don't know what the source of truth is for that,” Deutsch said. “It's important to have a data strategy that defines how data is created, and, if data is in multiple places that you trust, that it's the same information, defined in the same way, and built in the same way at the same time, so that the data is also timely, and you can look at data in multiple places and trust that it's the information that everybody is seeing.”

Torgerson added that data silos and their implications in healthcare could result from larger organizational issues.

“Most data silos are the result of organizational silos,” he said. “So, you can architect a data-siloed solution. One entity can put data in a bunch of different data servers.”

For instance, different healthcare organizations can hold various, incomplete, or duplicate pieces of a patient’s medical record, resulting in numerous challenges.

“When we talk about healthcare data, for example, your medical record and my medical record, who has the whole thing? My PCP may have a piece of it. My specialist may have another piece of it. My hospital system may have yet another piece of it. My health plan may have another piece of it,” Torgerson explained.

“They may have all similar information,” he continued. “Let's say they have blood pressure readings. Well, whose blood pressure reading is the right one? Well, they're all probably right. [But] which one's the source of truth? Well, I don't know. How do I understand that? Because they're all going to have that similar piece of information, but it might be different. They might have typed your name in differently in different places. So, now I've got a data quality issue.”

Accessing all this information from the consumer side presents a further challenge because it requires approaching each healthcare organization individually. Each may provide the data in different formats, making putting the information together in an efficient, meaningful way difficult, Torgerson stated.

These issues in data fluidity and tearing down data silos are the major drivers of using decentralized architecture networks in healthcare.

USE CASES, BENEFITS, AND CHALLENGES

There are myriad use cases for decentralized networks in healthcare, with breaking down EHR data-related silos being a significant one.

“Probably the biggest, most obvious [use case] that most people will be able to relate to is this idea that, ‘Man, wouldn't it be great if there was one universal understanding for me of my medical record and my medical history that I could access any time that I wanted, in any way I wanted, and that anybody that I wanted to could access in any way they needed to get to the things I want them to get to?’” Torgerson explained.

A similar use case is related to accessing provider information, which is relevant to patients, providers, payers, and other healthcare stakeholders.

“Where's the universal list [that outlines] what doctors exist and what they do? And what health networks are they a part of? Are they covered by my insurance plan?” Torgerson asked, indicating that this information could help healthcare organizations with credentialing activities for physicians, provide insights to develop and manage insurance plans, and guide patients in choosing appropriate healthcare services.

He noted that there are many use cases for a decentralized architecture network outside of these as well, some of which are common across healthcare organizations and domains.

“Where truth matters, that's a good candidate [for implementing a decentralized architecture network],” Torgerson stated.

However, Deutsch and Torgerson said there are technology- and adoption-related challenges to implementing decentralized networks in healthcare.

In terms of technological challenges, Deutsch noted that HIPAA compliance could create a significant workload for those looking to deploy a decentralized architecture network.

“Thinking about healthcare space, everything needs to be HIPAA compliant, and that takes a lot of work,” Deutsch explained. “It's challenging because every step of the way, especially if you think about all the different distributed systems that we would be working with, every one of those needs to be standardized, compliant, and then connected and working together.”

He also pointed out that gaining staff buy-in can also slow implementation.

“This is a new technology. People don't understand it. They might be afraid of it. They don't know what it is. They're used to doing things a certain way, and in the healthcare space, especially in the provider space, the providers don't have big technology budgets. They don't. If they have to buy a new piece of software or hardware to make something happen, that takes a lot of work. And you have a lot of hospitals and clinics that also need to do the same thing,” Deutsch stated.

According to Torgerson, trust and adoption are difficult to obtain despite the benefits of decentralized networks, like enabling data exchange across healthcare entities.

If the ultimate goal is to break down data silos across healthcare organizations, decentralized architecture networks would need to be widely adopted. Still, not every entity would want or be able to adopt the tech, Torgerson stated.

Another roadblock lies in getting organizations and stakeholders to trust that these networks are secure, which he illustrated through a cryptocurrency analogy.

“It's one thing to have a small crypto investment that's operating in that world. It's a whole different thing to having your entire life savings tied up in that world. And when we're talking about our medical records, it's the equivalent of our life savings in terms of our health information,” Torgerson said.

The wealth of data found in medical records makes them valuable, but it also makes them dangerous, he noted.

“So, if I'm not confident that [a decentralized network] is bulletproof in terms of security, and the stuff that's happening with crypto right now is not going to give people a lot of confidence [in] the technology... if you don't have trust in what's happening, it's going to be really hard for people to buy into the idea of doing this,” he said.

“Of course, in a HIPAA world, in the US in particular, explicit buy-in by the individual is going to be required for this to happen,” he continued. “And so, the benefits to me, an individual, are substantial, but the risks, they're substantial too. And the balance of those two things is really going to determine how realistic it is for these things to emerge in the healthcare space because you really got to manage these two very different headwinds in terms of the company, the corporate side participation, and then the individual and public side trust, to have these things.”

If trust can be gained, however, there are many potential benefits for improving healthcare.

“One of the other conceptual benefits here… is when you bring together data from a broader set of entities, your ability to generate meaningful insights that will improve the quality of care and improve the lives of people in our community actually starts to go up exponentially because everything you attach to it gets more powerful,” Torgerson said.

These improvements would provide a metric to measure decentralized architecture network deployment against, Deutsch and Torgerson explained.

Broadly speaking, the measures of success would be concerned with quality of life and care, Torgerson explained. Using the example of having all of the patient’s medical record information in one place, he indicated that these metrics would translate to a decreased likelihood that things will be missed in that patient’s care because of a lack of information, which will result in better outcomes overall.

“On the efficiency side, we spend an awful lot of energy trying to keep this stuff straight, particularly like provider data,” he continued. “If there was a solution where everybody didn't have to reinvent the wheel everywhere across healthcare, that improves efficiency. Which, by the way, if everybody improves efficiency, there is at least the opportunity that those benefits flow into the overall cost of care back to the individual.”

However, much work still needs to be done to effectively operationalize these metrics before the widespread adoption of decentralized architecture networks could deliver significant improvements across the healthcare sector, Deutsch warned.

“I think you’ve got to be careful, though,” he stated. “In our industry, a lot of these measures are lagging. How long does it take to show a population has improved their health or an individual has improved their health? And I think if you want to judge how effective something like this is, you need to give it some time.”