Data discussions were heard far and wide this past year. Most recently during the BIOMEDevice San Jose 2019 keynote panel discussion, “Digital Health: Beyond the Worried Well,” David Houlding, principal healthcare lead at Microsoft, encapsulated the topic nicely: “We have a very medical device centric perspective at BIOMEDevice San Jose, but increasingly we need to think about the solutions behind the medical devices, incorporating the cloud, artificial intelligence, machine learning,” he said. “It’s about the devices and about the data, but it’s really about deriving actionable insights near-real-time from that data that empower healthcare professionals to make better decisions to improve patient outcomes, to reduce the cost of healthcare, to engage patients better, and even to improve the experiences of healthcare professionals. But underpinning all this is that we have this highly sensitive data coming in—how do we ensure privacy, security, and compliance?
“It’s a complex challenge—we have a lot of different parts—the medical device, the Internet of medical things, devices in the hospital and clinic, implantables, wearables, devices in the patient home,” he continued. “These are all collecting data, and they have fantastic potential to yield actionable insights that can reveal an imminent episode and be used to avoid a stroke or heart attack. That has vast implications for preserving the quality of life of the patient and for reducing the cost of healthcare if you can avoid these kinds of episodes.”
Why Is Data Hot Now?
If you think about it, there really isn’t anything new about the need for medical device data. Medical claims and regulatory approvals have in some way typically depended upon data from a clinical trial, whether for the medical device referenced in a PMA or for the predicate device in a 510(k) application.
And yet, “data is in its renaissance,” Kevin Liang, senior director of medical device and diagnostics strategy at Veeva Systems, told MD+DI. Liang said that historically, medical device data has involved “small” groups, say of a few hundred patients. These patients are also relatively homogeneous in nature, having similar comorbid conditions and health profiles to meet the enrollment criteria for the studies that collected the data, he said. “But real-world data sets could be massive by increasing the patient population, and as importantly, the diversity of the population that we have data on,” he added. Given such size, “you might be able to recognize risk categories for smaller populations of patients as a way to enable decisions around personalized medicine” within those large data sets, he said.
And data can help with more than just marketing approval. “Data drives reimbursement,” Liang said: “For innovative products, it is hard to get paid so having data is key [for establishing] clinical efficacy and economic efficiency.”
Collecting performance data after marketing authorization can be challenging for any product marketer, but perhaps more so for medical device companies. “Previously they left it up to registries,” but medical device registries didn’t have quite the reach and maturity that pharmaceutical registries did,” Liang said. In addition, “EHR typically hasn’t included devices, so it is even harder for device companies.”
Europe’s new Medical Device Regulations may have in part fueled the recent push toward data. Veeva is offering a single-platform approach hosted in the cloud for collecting such data, explained Seth J. Goldenberg, PhD, vice president, Vault Medical Devices and Diagnostics, Veeva Systems. But he said that his company is helping “to build tools to move medtech beyond MDR. The new normal is more data and more connections across traditional business silos that all need visibility to this information.” Once collected, “it needs to be processed quickly so you can recognize trends,” he said.
Liang said that medical device companies are taking more responsibility for data collection today. And this is where the new era of digital health products could prove useful. “Connected devices could provide that data,” said Goldenberg. “The tools are there—you just need to connect disparate data sources.”
FDA is tackling the issue of “real-world evidence” and is working with medical device stakeholders on the National Evaluation System for health Technology (NEST), which the agency said will “generate evidence across the total product lifecycle of medical devices by strategically and systematically leveraging real-world evidence and applying advanced analytics to data tailored to the unique data needs and innovation cycles of medical devices.”
“Right now, NEST gets all of its data and licenses directly from hospitals,” explained Goldenberg. “When NEST matures to the point where they are bringing all of that data together into a common platform, Veeva can help—in fact, that is where we can help enable advanced data analytics, but the organization isn’t quite at that point yet.”
There are some very real challenges when trying to amass, handle, and gather data from multiple sources. “Payers will be critical about how data is put together, and they should be. There needs to be a better peer review process as well as the right experts looking at the data,” Goldenberg said. Data analysis also has to be “statistically powered.”
OK, So How Do You Get Data?
Increased medtech connectivity has certainly eased data collection.
While it may seem that the Internet of Things is “boom—all of a sudden here,” Bill Betten, president of Betten System Solutions LLC, described at MD&M Minneapolis 2019, he said that “it has been a long time coming. In the 1990s IoT really began with the transmission of radiological images and so there’s been a lot of work in that area, but the big thing that has changed is consumer technology moving into it [with] small wearable devices.” Betten moderated the MD&M Minneapolis Panel Discussion, “Top 5 Things You Need to Know about the Implantable Internet of Things."
Brian Chapman, partner and leader of ZS’s medtech practice of ZS, attributes today’s focus on data to the intersection of two important things:
- "A general recognition that understanding more and connecting actions with outcomes will provide feedback and understanding that will drive standards of care. This is not new, but as capabilities rise in data collection, aggregation, and synthesize rise, and coupled with machine learning, the promise of data in healthcare is becoming even more obvious.
- "Entrant of big, ostensibly ‘tech’players into healthcare has everyone realizing the potential. If you want to look for additional supporting trends, look no further than Google’s proposed acquisition of FitBit, which is about both hardware and, possibly more importantly, access to an enormous trove of data. More proof in the news cycle recently is the visibility that just emerged about the relationship between Google and Ascension, which is all about data."
Teresa Prego, senior vice president of marketing and business development, explained the connection between digital health and data during her keynote talk at MD&M Minneapolis. “How do you even define digital health? [It’s] taking technologies, information management, and advances in pulling information about the patient and applying those technologies to healthcare,” she said. As an example, she explained the PKG [Personal KinetiGraph] "watch" her firm is marketing for tracking the progression of Parkinson’s. The device collects accelerometer data and creates a report that can be provided to clinicians. It offers a “7-day wear without removing it for a charge, which you can’t do with an Apple Watch,” she said.
For such data-collecting products, you have to consider “the impact on health economics,” said Prego. For “uncontrolled Parkinson’s, it is $52 billion dollars a year,” she said.
One of the big challenges, of course, is identifying what data to collect. “The big challenge with IoT is getting useful data. It is identifying what data is out there that can be useful for diagnosis and treatment,” Mark Wehde, section head, Mayo Clinic’s Division of Engineering, said during the MD&M Minneapolis Panel Discussion, Top 5 Things You Need to Know about the Implantable Internet of Things. “Our focus is always on trying to identify information that our physicians can use to drive patient care and figuring out ways to capture that data.”
During The Medtech Conference 2019 panel discussion, “How Data Will Transform Patient Outcomes, Medtech, and the Practice of Health, and Why It Has Happened Yet,” moderator Brian Chapman asked his panelists about the difference between data and evidence and about some of the barriers and risks.
Richard Loomis, chief informatics officer, clinical solutions, Elsevier (which he said is transforming to become an analytics company and had acquired Via Oncology) said it “begins with having access to data. Historically, we’ve been challenged to aggregate and assimilate data to get to evidence.” Now, however, he’s encouraged by the influence of the 21st Century CURES Act. “We are finally at a tipping point where we can bring all required data together, understand it, be confident in it, and ultimately generate evidence from it,” he said. He expects the regulatory side to catch up to be able to use real-world evidence.
To “bring data together,” Loomis added, “you can’t go to one single source for the data you need.” He said that “forming partnerships” is necessary “to create rich datasets to unlock insights. We are working to get as close to patients as possible and interact directly with the care team, providers, and patients.”
Yvonne Bokelmann, president and general manager, strategic and value-based solutions for Zimmer Biomet, said during the discussion that “it’s coming up with the right structure in an organization and then convincing and collaborating with providers to ensure we can share data at an aggregate level on how to [collect] and contain data and safeguard it.” The other issue is determining how to integrate data from other places, such as robots and home care devices, she said.
Bokelmann added that Zimmer Biomet is “fortunate to have a relationship with Apple” with its MyMobility app for the AppleWatch for orthopedic surgery patients. Zimmer’s large randomized control trial of MyMobility aims to measure outcomes and cost, she said, and they are looking at “how do we help providers and patients get through the continuum of care in the most cost-effective manner?” With MyMobility, Zimmer Biomet is working preoperatively with total joint patients to educate them all the way through surgery and then to manage and deliver care at home along with physical therapy, she said. “We think we’ll see an evolution in the standard of care and reduced variability and cost,” she added.
After a couple years into Zimmer’s relationship with Apple, Bokelmann said “it is interesting to see opportunities,” she said, adding that the company was “interested in another mechanism to collect data beyond a traditional clinical study, physician inputs, etc.” She also described a “sea change going from an implant manufacturer to a solutions provider.” In addition to expanding data-handling capabilities, Zimmer Biomet acquired gaming expertise and now has data scientists on staff, she said.
Performance data could be in demand as healthcare models evolve. Loomis of Elesevir said healthcare delivery systems “are really challenged to participate in a value-based world. . . for some, as much as 50% of [care] is in some form of a value-based program. They are looking to the industry and to us for help. Reducing variability is critical to being able to deliver consistent high-quality care. . . .You have to have an understanding of cost and subsequent quality, so you need data to measure it.”
Chapman later explained the rising importance of post-market data to MD+DI. “There are two reasons for this. Post-market data is important first because it offers the opportunity to ‘prove’ the benefit of an intervention, which has previously eluded us. So many interventions lack a pure ‘randomized control trial’ to prove their benefit but using post market data allows a more clear benefit of value.”
Secondly, “while FDA approval is becoming marginally easier, reimbursement has actually become more difficult,” he continued. “The traditional view has been that once FDA approves a device, reimbursement soon follows. More recently we have seen examples of devices that are approved without reimbursement. This is such an important development that sometimes start-ups will refer to this as “the valley of death” between approval and reimbursement where many start-ups have died.”
Data may also materialize from—and perhaps transform—traditional procedures such as surgery. In the “Future of Surgery” panel discussion at The Medtech Conference, Joseph DeVivo, chief executive officer of InTouch Health, told the audience that “what you’re going to be hearing a lot at this conference and a lot going forward is using data. . . . What the industry is waking up to is that there’s a lot of performance data above and beyond the actual clinical data that determine clinical success.” For instance, “one of the most exciting things I believe that will occur in surgery . . . is connecting all these OR environments to be more intelligent and to be able to pull data and to do machine learning and AI and make things easier for surgeons.”
Copanelist Nancy Briefs, cofounder, president, and CEO of AltrixBio, agreed, adding that “Data in surgical suites as well as throughout the hospital is really going to be very important for providers, healthcare systems, and payers, so getting data that is actionable and useful is really key.”
DeVivo also believes that there is a need to connect ORs given the need for clinical specialists even if they’re not physically there. It involves “connecting the ORs and giving them more power to push and pull data real-time to physicians. We’ve done some of this already through telementoring and teleproctoring surgeries today. . . We looked at this platform to continue to allow for efficient and clinically necessary ways to deliver intelligence into the operating room.”
There’s also hope that such data may also help address one of healthcare’s biggest problems—cost. “The global economy can’t keep paying for healthcare,” said DeVivo. “Innovation has to be the solution. Virtual care is one of those panaceas because the biggest cost in healthcare is labor. . . . If you have one hour of a physician’s time and it’s not used properly, you’re never going to recapture it again.” What is needed is “the ability of looking at ways to not just use clinical data but performance data and preop data and then to allow for world-class skills sets to not just be in world-class environments but to be in every environment and democratize it,” he added. It also involves “allowing people to provide that care, by streaming data to make sure we are making the right decisions. Robotics will be the equalizer around surgical complications, time, and excellence. It will all be around empowering physicians to make better decisions.”
What About the Challenges?
Chapman outlined the following challenges to collecting data:
- "Privacy: the risk of a breach is significant. Of course, there is also fairly strong legislation covering this space as well.
- "Much of the data specific to the site of care. Once a patient leaves the hospital, the facts about what was done and what the outcome was is much harder to combine.
- "A lack of interoperability. This is responsible for some of the issues around tracking across sites and episodes of care and is even a problem when different hospital networks don’t share data, or insurance companies don’t share. The full history of a patient doesn’t follow that patient over time because they change payers or move through the healthcare continuum."
When it comes to data, privacy, security, and even GDPR, it is important to not only adhere to regulations, but also to be able “to defend our position,” said Tammy Mae Moga, principal global privacy and security at Medtronic, who spoke in the MD&M Minneapolis panel discussion, “Building Better Cross-Functional Teams for the Digital Health Explosion.”
“It’s a device with information, but what other uses will it be used for?” asked Moga. She also pointed out a number of other questions to consider, such as whether data can be used that is truly de-identified? Also, for updates or a patch, does the device need to be brought to the manufacturer? Can a patch or an update be handled automatically, through an app on a phone? What teams need to be involved to make that happen?
She said that “because we have a lot of data involved . . . now at the table in the beginning stages we incorporate privacy or security . . . create fundamental specs or requirements so that designs adhere to regulations. . . Which ones do we have to follow and what do we need to do?” Also, “Who would be adversely impacted if something were to happen? Think about all the adverse things that could happen and then you know who should be at the table.”
During his MD&M Minneapolis panel discussion, Mayo Clinic’s Wehde said that “the security issue is a big deal with implantable devices and wearable devices. It is not unusual for people to try to hack these devices—this has been demonstrated since 2008,” he said.
Also, Wehde said that “we are all used to WiFi and Bluetooth giving us a lot of bandwidth. But these things don’t work very well for implantable devices—the frequencies they use don’t transmit through skin very well.” There are also other issues with WiFi in terms of security. “We are looking at custom RF transceivers and frequencies that give us good transmission through skin, but these are lower frequencies and have bandwidth limitations, which is one of the reasons we need smarts inside our devices.” He pointed to IEEE WBAN standard, MICS 400 Hz. “There are huge tradeoffs in terms of the amount of data we can get out versus the frequency versus how much power we need versus how close we have to be. We don’t want to implant a device with a receiver mounted right outside the skin—we want to be able to communicate a little farther than that.”
Another challenge is what Betten called “data fusion,” acknowledging that data is often “siloed.” He asked his panelists about their vision on how data could get pulled together. Answered Brian Kronstedt, senior product development manager for Preventice Solutions: “As we think of data fusion, sure, we have a dashboard that can pull together and show data, but the collective we need a next-level algorithm that says an individual has AFib, weight gain, lower O2 saturation, they may be a candidate for congestive heart failure. I think the market has started to trend this way and it will continue to build out these capabilities and take several different inputs that themselves were established by algorithms and data science. So it’s coming, albeit not easily. . . . Just getting data sets for the all-important verification and validation of such capabilities is pretty difficult because you need definitive inputs to demonstrate that these flags for things like congestive heart failure are functioning well.”
Chapman told MD+DI that there is a need for collecting data from multiple devices and sources and assemble them for insights into the condition of an individual patient insights. “There is absolutely a need for this. If you can combine big data with population health, there is huge value to patient outcomes, standards of care and costs by getting the right therapy at the right time to the right patient. It cuts across conditions. Some of the ‘easier’ applications come in oncology – genetic sequencing, and clear documentation of the treatments and the outcomes will all lead to vastly improved protocols. It is clear that there will be value. The question for me is not about the patient value in solving the privacy, interoperability and ownership challenges. The only real question for me is about the path, the duration of the journey, and the winners and losers along the way.”
“There are several opportunities and challenges, but if I were going to pick one out it would be good ole fashioned power management onboard these devices,” said Kronstedt of Preventice Solutions. “As they get smaller and more capable, they get more bells and whistles, algorithms, and elegant UI—that takes battery. People rightfully expect elegant usable devices, and the size and weight optimization is a challenge.”
“As we monitor thousands of patients simultaneously, that’s a huge amount of data that Preventice is responsible for,” Kronstedt added later during the discussion. “We do our best to depict the most interesting content—the most interesting ECG—and get that up for assessment as soon as possible. . . . A lot of the ECG that is collected is not interesting—it is just normal sinus rhythm.” An algorithm highlights the most interesting pieces, he said, and up in the cloud are “e-neural networks and more sophisticated algorithms that require more horsepower to manage the data and make definitive reports. Even the regulatory aspects of that are a bit burdensome. We have to maintain and utilize datasets to verify and validate these types of sophisticated algorithms.”
Wehde said his team wants to create a platform that can be configurable and customizable to be able to handle data from different sources. “We might want to take ECG data, provide some stimulation for doing work with epilepsy, and monitor the vagus nerve at the same time. There are different things we want to be able to try to do for which we cannot buy an off-the-shelf device. So, we started this effort to put together an implantable platform.”
Wehde detailed the components that should be considered for such an implant:
- A sealed can that can be put inside the body, which is a bit of challenge.
- A way to get signals and sensors in and out.
- A power source and power management.
- Very small, low-power microcontrollers and FPGA circuits.
Ultimately, though, it all comes back to data and data management. “One of the challenges is the huge amount of data you get. If you are going to monitor someone’s ECG continuously, that is a huge amount of data and no physician can sit there and look at that,” he said. “So, you need algorithms back in the cloud and analytics to help reduce that dataset. But you still need to transmit that data to the cloud. We are looking at making sure we have the power inside our device to be able to add some AI to it so we only send data that needs to be sent.”
AdvaMed Steps In
AdvaMed launched its Center for Digital Health in October. Digital Health is a horizontal affecting all companies, said Andy Fish, chief strategy officer, AdvaMed, during a conference call on the new center. “There are new and emerging issues that need work on our part, such as data privacy and the California Consumer Privacy Act.
When MD+DI asked whether AdvaMed has a definition of digital health, Fish said that “We don’t use a formal or rigid definition but generally consider digital health not as a health care vertical but rather as an evolving set of data-generating technologies, data-driven technologies, and data uses that cut across medtech and health care to improve patient outcomes and health care quality and delivery.”
He added that AdvaMed expects “the term ‘digital health’ to gradually become less meaningful as data generation and use become more embedded in health care, to the benefit of patients. While medical technology manufacturers should not chase data just to wear a digital health halo, in general there are significant opportunities to improve patient care by bundling data-driven insights with specific technologies.”
During the conference call, Fish said that “there is a recognition that this is an area of intersecting issues affecting multiple companies. We are seeing technology evolution with a data revolution.” He pointed to continuous glucose monitors as an example, saying that “data is an integral part of that application.”
“Use of data is facilitating the assessment of outcomes based care,” he added. Fish.
AdvaMed has had workstreams related to digital health for some time, but now they will be handled through the center, explained Zach Rothstein, vice president for technology & regulatory affairs. These include projects surrounding FDA’s re-issuance of clinical decision support software guidance and cybersecurity for medical devices.
In addition, “We are pushing for tangible outcomes with FDA, such as AI regulation and precertification concepts,” added Fish. “We are considering new benefit categories that would have to be enacted by Congress.”
For instance, the center is kicking off a digital technology payment workgroup by working on a whitepaper on barriers that make it difficult for individual technologies to be covered and on policy recommendations that suggest pathways, explained Richard Price, Senior Vice President, Payment and Health Care Delivery Policy. “It will include a scan of the landscape of digital technologies and categorize them into technologies that require direct coverage for digital health versus indirect coverage for those technologies that already improve efficiency, patient care outcomes, and quality,” he said.
“Most of our work has been focused on Medicare issues, but we will consider private coverage issues. [Today] to get coverage you have to fit into an existing benefit. There [haven’t been] changes in the basic benefit structure since telehealth coverage dating back to 1977,” Price explained. “We are hoping the whitepaper project will identify specific Medicare regulations drafted for other areas that create barriers for digital health and to make changes to the existing categories to remove those barriers.”
And when asked whether companies developing wellness products not medical devices should consider joining AdvaMed for access to the new center, Fish said that “The work we are doing at AdvaMed, including under our Center for Digital Health, is highly relevant totechnology manufacturers focused on broader wellness applications, especially with regard to reimbursement, value assessment, data privacy and cybersecurity.”