Altering the Ecosystem

By evaluating how Big Data is used in other industries, the imaging community can find solutions to complex challenges and help modernize healthcare. 

By Bipin Thomas

A new healthcare ecosystem is emerging enabled by consumer access to health information that can be made ubiquitous and shared in creative and astounding ways. Stimulated by advanced connectivity and access to banking, investment, transportation, and other information, consumer expectations are rising. Consumers are also becoming more discerning shoppers of healthcare services, especially as ever-increasing out-of-pocket expenses make them more vigilant. As a result, consumers will demand more information and services from the healthcare value chain. This phenomenon is driving a larger business model shift from business-to-consumer (b2c) to consumer-to-business (c2b), and there are powerful incentives for payors and providers to quickly adapt to this new reality. This ecosystem will include makers and users of medical imaging systems. All will play a significant role in the connected continuum of care.

Under this new construct, members of a new healthcare ecosystem are drawing closer together with the advent of smarter medical technologies and real-time information sharing. The data made possible by wearable, wireless medical devices—and increasing awareness about how those data can be used—are helping healthcare providers, home health agencies, pharmacies, and others to embrace a “Big Data”-driven form of healthcare. This new ecosystem takes a 360-degree view of consumers, seeking to meet more of their needs on their terms.

As the focus on the individual consumer tightens, healthcare has much to learn from the industries that have already made this shift. Just as the auto, consumer electronics, and telecommunications industries have found ways to drive out excess costs by focusing more on the features most valued by their customers, so must the imaging industry. The winners will be those who can deliver exactly what the customer wants—nothing less, nothing more—at the best possible price.

In healthcare, of course, one of the most difficult issues is identifying who the “customer” is, and who are the critical stakeholders. In terms of medical devices, physicians might want to use a device or technique that evidence-based data do not justify, a situation exacerbated by the dearth of comparative effectiveness reports. To be selected for use, a device has to earn approval not only from physicians, but throughout the ecosystem: it must be approved by a national or regional authority, chosen by the provider community at large, endorsed by the highest volume users, and, ultimately, accepted by the patients themselves.

With this complexity comes opportunity. The unique nature of healthcare consumers—those whose needs are influenced not only by their financial standing and preference but also by biology, genetics, and wellness—could ultimately put savvy healthcare organizations and medical disciplines at the cutting edge, rather than the tail end, of the Big Data revolution. David Whitehouse, chief medical officer of UST Global, predicts, “We will use big data to analyze the needs of the whole population, incorporate comparative effectiveness research, understand from unstructured data how imaging supports behavior change and improved decision-making, and be able to predict the match between what data is truly necessary to diagnose, to treat, and to track.”

Three Categories of Intelligent Devices

Medical devices were by and large stand-alone machines focused on a specific task—and on that task alone. Any information they contained had to be extracted at great effort. If analysis of these data was even on the table, it required a painstaking rekeying of information into a separate database. But now, with standards of interoperability in place, devices from personal glucose monitors to MRI machines can send a patient’s vital signs and other biomarkers directly and automatically to a central repository, or even to the cloud, to be analyzed. In some sophisticated settings, these devices can “talk” to one another as well.

Intelligent medical devices come in three categories: small, medium-sized, and large.

 Small

This category includes lightweight, miniature, wearable devices that can monitor everything from heart rate and pulse to real-time glucose levels and long-term cardiac activity. These devices boast enhanced features such as time-stamping and real-time biomarker monitoring, but are designed to be easy to use. Automating the stream of data sent to a consumer’s personal computer or to the cloud underscores this ease of use: as these technologies have become more compact and more features-driven, their information can be sent directly to a central location for data or seamlessly added to a patient’s electronic health record (EHR). These repositories offer responsible healthcare organizations great opportunities to improve care and streamline business practices.

Further, real-time capturing of vitals and biomarkers helps create a holistic view of the consumer. This category of medical device thus yields truly functional data, not just historical data or a record of the past.

Medium-Sized

Medium devices are those devices such as point-of-care ultrasound machines that have been mobile for a number of years now. As medical manufacturers have focused their energies on increased portability, many technologies have migrated from hospitals and clinics to hybrid locations such as pharmacies (Walgreens, CVS) or even mega-grocery stores. Consumers today are less startled to see medical scanners in the same place they do the household shopping—and soon enough may expect to get an ECG reading while their family selects eggs.

Large

Stationary and facility-based, the devices in this category include bread-and-butter imaging technology such as PET and MRI machines. While this category lacks the mobility of the other categories, machines here are also making major strides in data collection and communication. For example, new analyses of imaging data are helping map the borders and growth of tumors, and then delivering precise doses of chemotherapy or radiation within those borders.  The advent of 3D technology like tomosynthesis will provide renderings that are highly accurate, have less false positives, and are able to pinpoint abnormalities perhaps in the future down to the cellular level.  Retrospective and real-time data will leverage their utility both clinically and economically across the continuum.

With new ways of producing and sending data in each of these categories—and with faster and more powerful ways to process and interpret that data—radiologists and imaging professionals need to start asking the right questions of this information. By mining these data to improve work practices, methods, and processes, thereby improving patient outcomes, radiology can become a more valuable contributor to a more valuable healthcare system.

Role of Imaging in the New Ecosystem

Before it can contribute to the new ecosystem, imaging has to travel a learning curve. What exactly can the diagnostic imaging community gain from looking at how Big Data is used outside of healthcare? Solutions to problems range from the simple (equipment failure, excess inventory) to the complex (service models that lag behind consumer expectation).

Whitehouse asserts, “The great imaging departments are creating the future, not just keeping up with it.” As a discipline and an industry, imaging should follow several important lessons from other industries about Big Data—and avoid certain pitfalls as well. What’s most exciting is the prospect that imaging could take the lead in modeling a new form of healthcare.

Wisdom from Manufacturing

Large equipment on manufacturing shop floors is pumping out endless streams of data. The real-time synchronization of data from shop floor to top floor systems delivers better insights on equipment utilization and supply chain visibility. Big Data analytics have become a must for manufacturers interested in reducing machine shutdowns and improving performance. Shop floor devices now regularly feature real-time collection of data, which ultimately enables managers to anticipate problems before they occur. Because the machine data are sent directly to the cloud, where analysis is automated, managers can see in real time when preventative maintenance is necessary. The same process can be adapted in medical devices through real-time synchronization of their data with hospital management systems and running advanced analytics on the cloud.

Some companies, however, get so caught up focusing on real-time problems that they neglect to think strategically. In designing the algorithms that analyze machine data, engineers should think more broadly about the eventual application of the information. Stepping back to take advantage of the big picture—not just whether this machine is about to fail right now, but what are its typical failure points? Can those weaknesses be fixed?—is a step that many companies skip, to their detriment.

Thinking Ahead: New Imaging Technology

Imaging has made particular strides in capitalizing on real-time data with broader and more long-term improvements in mind.

For example, the MITA Smart Dose for Computed Tomography (CT) standard, an initiative of a collective of medical device manufacturers, identifies four key CT equipment features that enable optimization of radiation dose delivery without compromising the quality images necessary for an accurate diagnosis. Automatic Exposure Control and Pediatric and Adult Reference Protocols help the CT equipment operator better tailor the amount of radiation dose received by the patient to the specific exam and clinical question, and Dose Check technology informs equipment operators when an equipment setting would likely exceed established dose thresholds. Both measures allow clinicians to confirm correct settings prior to a scan and avoid unnecessarily high levels of radiation dose to the patient.

As is typical of Big Data, the MITA Smart initiative has ramifications beyond the single consumer. The technology’s DICOM Dose Structured Reporting is used by imaging systems to communicate data between the scanner and any connected computer system, sending dose information to the patient record and to the facility itself, helping improve quality assurance measures. More broadly, the systematic collection of dose information from CT equipment across the country ultimately makes possible the establishment of national dose reference levels.

In terms of smart imaging, it will only be a matter of time before capabilities in feature enhancement and image processing will be done automatically, enabling the automatic classification of images. The industry is headed toward machines and devices that adjust their view, processes, and imaging presentation based on immediate feedback as they intelligently interpret data and tailor their approach to the goal at hand.

Thinking Ahead: New Business Configurations

As imaging capabilities increase, the capital outlay essential to make those capacities available increases as well. Similarly, as sophistication increases, so does the complexity of repair and maintenance agreements. Downtime monitoring and the reliability of service agreements, predictive software that can forewarn of system failures and increase prophylactic intervention rather than suffer loss of capabilities, are increasingly crucial. In fact, new approaches to leasing suggest that payment should be for the guaranteed availability of the capability with the burden for repair, replacement, and updating borne by the imaging manufacturer.

Remote operation is another way imaging centers might capitalize on new technological capabilities and data-sharing: nearby healthcare providers could share MRI machines instead of each buying one, and the resulting images and other clinical information could be sent to the correct clinic wirelessly. Both of these shifts—in service burden and in remote operation—would have major business consequences for everyone in the healthcare ecosystem.

From Rolls Royce to the Emergency Department

Lessons from other industries can come not just from the raw data, but from how they process that data.

Consider the Visensia index, for example, a development drawn from algorithms created by UST Global partner company Oxford Bio Signals (OBS). OBS had been working with Rolls Royce to create improved engine monitoring, including early warnings of potential negative events by leveraging algorithms that worked by analyzing the data from multiple continuous monitoring devices in real time (“data fusion”). Part of the 2006 Trent 900 development program for the Airbus A380, this multisensor data fusion system “learns” the characteristics of normal engine operation and instantly recognizes any variances from this norm. As Rolls Royce says, “The aim of a monitoring system is to maximize reliability and availability. A monitoring system will not stop a system from malfunctioning, but will log system data from which system characteristics can be deduced. This provides our customers with advanced data on the status of their system and data to plan maintenance schedules around.”

By applying those same principles to vital sign monitoring, OBS looked to take the current system of single parameter vital sign monitoring used in ICUs and CCUs and utilize data fusion algorithms to create a similar system capable of recognizing variances from the norm that would create an early warning system of critical life-threatening events.

The result is the only FDA-approved early warning system algorithm, which responds to patient deterioration an average of 6.3 hours earlier in the case of critical instability. Not only did the Visensia monitoring result in a 60% reduction in serious instability, there were no unexpected deaths during the 8 weeks of Visensia monitoring—in contrast to six unexpected deaths in the 8 prior weeks. In 24 months of Visensia monitoring, no patient had an unexpected fatal cardiac event. Further, the system lessened alarm fatigue substantially, with a 79.3% sensitivity and 80.5% specificity, and only 1.6 false alerts per 100 hours of monitoring.

Imagining the Future of Imaging

Big Data offers real opportunities to the imaging community beyond mere modernization; it opens up a whole new vision of “imaging” itself—one that spans from biological systems to delivery systems, from nano-capabilities to large data mapping, from advances in optics and material sciences to how, when, and why we use visual data. In the new healthcare ecosystem, imaging should take the lead in deducing where knowledge provides wisdom and a different direction—and where it just provides more data.

Considering the advances that other industries have been leveraging for some time now, we might ask how imaging centers and practices can use data to change the way healthcare is conducted—in terms of not only clinical treatments and better etiology, but also resource allocation and business practices?

First, imaging will have to push systems analysis beyond simple metrics in order to ask critical questions. These questions range from the immediate—how many MRIs does a 300-bed hospital need to understand the impact of maintenance/repair life-use patterns on imaging quality by manufacturer?—to the profound—how does imaging support behavior change and improved decision-making?

Second, imaging will have to manage data to determine appropriate financing and service agreements, and will have to identify the outcomes metrics (clinical, financial, and operational) that foster an environment of both innovation and sustainability.

Ultimately, the industry’s focus has to be on determining what is the right level of information necessary for the clinical task at hand, whether that is diagnosis or intervention. The inappropriate use of expensive but unnecessary imaging capabilities beyond what is necessary is a major cause of medical waste. New capabilities will have to be put through much more rigorous comparative effectiveness studies to determine the subtle differences where a particular methodology suggests enhanced approaches from the start or when less sophisticated approaches done first would be wiser and produce as good if not better outcomes more cost-effectively. In this capacity, Big Data must grapple with the thornier issues of globalized healthcare: what level of visual definition is necessary where, and for what purpose?

For all of the challenges and wrinkles that will need ironing out, smart imaging devices, which integrate and output Big Data, offer great clinical, operational, and consumer promise. The convergence of such devices and data will respond to and be part of the revolution of rising consumer and outcomes expectations. The imaging industry is reimagining what it can do and moving quickly toward convergence through Big Data. Consider robotic surgery smart 3-D imaging devices that will soon have real-time conversations with robotic surgery arms, improving procedure depth and accuracy, creating operational efficiencies previously only dreamed about, and coordinating diagnosis and treatment as never before. This and other smart imaging devices also will offer physicians powerful visual tools to educate and engage consumers to the extent that they will become co-stewards of their care, the final and critical continuum link.

###

Bipin Thomas is President-Health Group, UST Global.