Next-gen analytics: Here's what's coming in the future

Hospitals should expect orders of magnitude more data – but will also see emerging tools such as artificial intelligence and 5G connectivity helping to put both structured and unstructured information to work.
By Bill Siwicki
09:11 AM
Share
next generation analytics reviewed by doctor on EHR

The healthcare analytics market is booming and will be worth close to $54 billion worldwide by 2025, according to a March 2018 report from Grand View Research.

Given the need to achieve the Triple Aim, along with the rise of precision medicine and the move toward value-based care, data analytics have never been more important to healthcare provider organizations.

As the technology continues to grow and mature, here's the pressing question for healthcare and IT leaders: How will analytics tools evolve – and what should they expect to come next?

We asked experts from across the industry about major shifts on the horizon. Here's what they said healthcare decision-makers should be tracking now.

We're getting closer to a 360-degree view

Healthcare providers have long sought the elusive 360-degree view of the patient, and soon the healthcare industry will get it, said Brandon Purcell, a senior analyst at Forrester Research who specializes in analytics.

"Because analytical models and artificial intelligence are only as good as the data used to teach them, healthcare providers will make significant investments in the creation of foundational data assets that link patient data from disparate sources," Purcell said. "The completeness and quality of this 'single source of truth' will be the key to differentiation for healthcare providers, enabling them to provide proactive care, personalize services and reduce operational costs."

Purcell also said that unstructured information will become more commonly used.

"Text, image, audio and video data have long been analyzed in a vacuum, typically by a human being," Purcell explained. "Solutions like Watson Health that are able to diagnose results from medical images are just the start of a trend in healthcare toward using deep learning to analyze unstructured data."

Technology now allows for conversations with patients – whether in person or on the phone – along with text from emails and SMS messages and all types of image and video data now can be analyzed by neural networks. The results of this analysis: structured data points that can be added into the holistic view of the patient.

"Healthcare providers should look to adopt solutions that have been trained in their specific use case and offer the ability to further customize models through the process of 'transfer learning,'" Purcell added.

Longitudinal records are a foundation

Healthcare is an industry rich with data, yet many provider organizations continue to focus on single vertical views of that information – for example, analyzing outcomes at just one site of service or thinking about data in terms of individual care encounters.

Instead, health systems must look to analytics technology to provide a holistic view of the patient experience over time and across all sites and episodes of care, said Garri Garrison, RN, vice president of performance management at 3M Health Information Systems, which develops analytics and other healthcare technologies.

"These analytics tools exist today, but their success depends on the creation of the longitudinal record," Garrison said. "This is the next step in the evolution of analytics technology. To take advantage of new analytic tools, healthcare organizations must have access to patient-centric records that encompass the full spectrum of clinical care."

By applying advanced analytics to aggregated data from across all visits, episodes of care and patient populations – and evaluating it against key performance measures – organizations can identify inefficiencies and uncover interdependencies between sites of care and between caregivers that may be causing poor outcomes or high costs, she said.

"With advancements in machine learning and artificial intelligence, we'll see a transition from descriptive analytics to predictive analytics," she added. "Integrating machine learning or AI with risk stratification methodologies will create a new set of analytic tools that support real-time interventions in care delivery."

Robust analytic capabilities will also advance performance measurement, providing meaningful information that can be used to promote real and sustainable improvements in healthcare quality and cost – especially as analytic tools incorporate social determinants of health to provide a more complete understanding of patient populations, Garrison said.

Emerging technologies beget more data

Jennifer Esposito, worldwide general manager for health and life sciences at Intel, points to some technological shifts that will affect analytics that healthcare CIOs need to understand.

"We are starting to see convergence in High Performance Compute, or HPC, and AI workloads," Esposito explained. "As data sets get larger and more complex, the lines between scientific computing and analytics are blurring. HPC clusters are well-suited to the increasingly parallel nature employed by AI-enabled analytics. An HPC approach to analytics drives additional architectural considerations such as high-bandwidth interconnect fabric between CPU nodes and the use of parallel file storage systems."

On another technological front, healthcare provider organizations will need to be planning for how analytics strategies change as the world becomes connected by 5G, she said. The move to deliver more care outside of the hospital is a trend seen worldwide, and 5G is going to open amazing opportunities to deliver new services and reimagine how patients are engaged, she added – adding even more information that can be analyzed.

"Examples of this could be immersive real-time virtual reality rehabilitation sessions in the home, virtual visits augmented by complex sensor data or field robotics controlled by centralized specialists," she explained. "Clearly, it will still be some time before 5G becomes a widely used platform. But it is coming, and it has the potential to generate orders of magnitude more data than what is being generated today."

Lesson learned: Data governance matters

In the end, it is important to remember that the first step to driving forward an analytic strategy is to identify potential sources of data and information available to an organization. Data acquisition is a laborious effort within healthcare due to its system designs, but necessary to have a complete picture of care and the opportunity to improve performance.

"We have been on this journey for approximately 18 months and have connected a few hundred data sources – EHRs, billing files, claims – all while knowing we have just begun to scratch the surface," said Derek Novak, division vice president and COO at Iowa's Mercy Accountable Care Organization. "However, we already are starting to see benefits of this work as we have created a uniform dataset that incorporates all aspects of patient, payer and provider information."

One mistake the ACO made early on was to jump right into analytics – only to later discover that the quality and consistency of the data was problematic.

"Pushing bad data out to executives and providers ultimately leads to a lack of trust and a lot of wasted energy for the organization," Novak said. "Having an analytics platform partner to work through the process to ensure a correct aggregation of data and ensure quality data is a crucial step in the process."

Novak said that providers looking to the future and getting into analytics should select a platform from a vendor that is willing and able to work with their data sources and has the ability to grow with their needs.

The platform should be able to work with all types of systems and users," he said. "It can't be yet another closed-off system with no access. Your analysts and other applications need to be able to access the data so you can actually scale for your organization."

For Mercy, the work enabled it to incorporate data across six regional chapters, spread out in over 400 locations and 3,500 providers. It afforded Mercy the ability to scale to 18 value-based arrangements and 310,000 members.

"While we didn't start with this number of lives when we started our engagement, over time as our needs grew, we needed to be sure that our data platform had the ability to grow with us under an economically sustainable model," Novak said.  

The big (data) takeaway

As analytics products continue advancing from descriptive to predictive and, in turn, to prescriptive functionality, hospitals will have more tools – including but not limited to technologies such as artificial intelligence, high performance computing and 5G – to more effectively use as much data as they can to establish longitudinal records with a 360-degree view of the patient. 

That will mean connectivity to literally hundreds of data sources, including EHR, text, imaging, audio, billing and claims, just to name a few. 

And yes, that will require sizable financial investments and heavy lifting to make sure the technology infrastructure is in place. But the opportunities for return on investment, through new revenue and cost-savings, are significant too. 

Twitter: @SiwickiHealthIT
Email the writer: bill.siwicki@himssmedia.com