Trial Shows Benefits to Improved Remote Monitoring
By Matt Phillion
A new clinical trial on remote monitoring found that while all patients involved saw improved survival rates, the type of remote monitoring had a significant impact.
Implicity, a developer of alert-based remote monitoring solutions, in collaboration with the Health Data Hub, looked at a database of over 68,000 patients linking real-world data from patients with cardiac-implantable electronic devices (CIED) to remote monitoring methods and compared mortality rates, annual hospitalizations, and the cumulative duration of hospital stays.
The results demonstrated that alert-based monitoring using Implicity’s platform was associated with greater performance compared to historical manufacturer solutions.
“We always say remote monitoring is better, but the reality is when it comes to science and evidence, there are not that many examples of proven clinical evidence of better outcomes with remote monitoring,” says Arnaud Rosier, CEO of Implicity.
Remote monitoring of cardiac implants (such as connected pacemakers) is a niche area, Rosier says, and thus there are few studies looking at how remote monitoring for cardiac care impacts mortality, rehospitalization, and cost reduction.
“As a physician, I and my colleagues know the value of remote monitoring. If you’re using a pacemaker and not being remotely monitored, your care is probably not being handled well by your physician,” says Rosier.
The study was in part intended to bridge the gap in proving the value of this type of monitoring. It wanted to examine not just the monitoring itself, but how the software used for that monitoring played into patient outcomes.
One of the reasons to look at remote monitoring for cardiac care now is the opportunity to decrease the burden. “Fifteen years ago, it was a burden—we knew it was beneficial, but it was such a pain to manage all the information coming through. The industry wasn’t organized to triage all the data coming in,” says Rosier. “It’s not what we were trained to do as physicians.”
This is where advancement in technology shines through, he explains. “Technology is needed to scale this. You don’t have teams of 20 or 30 doctors or nurses to triage the data,” says Rosier.
Triaging data offers an opportunity to identify risks in patients—and higher-risk patients in general—based on remote monitoring. Rosier describes an example of a patient he encountered years ago who had recently had a stroke. “I analyzed his pacemaker data and saw this patient had the data saying he could have a stroke the next day,” says Rosier. “If you weren’t having your patient remotely monitored, I realized, you weren’t fully doing your job.”
A challenge early on, though—and one that persists—is that a patient who is remotely monitored is probably being monitored better by default.
“People doing remote monitoring are better managing their patients than those who don’t, so the results are biased,” says Rosier. “It’s different than a drug trial. With remote monitoring, you have a control group with which you would do nothing, and then you have an active group who is very well managed; clinicians get money to do it, and there are resources to do it well. So basically, remote monitoring done well has benefits for sure, but in real life, people don’t have the same resources, and so no one knows the real impact.”
This new study, however, looked at a complete set of data in France to avoid the biases of a controlled trial. It showed less clinical impact from remote monitoring compared to controlled trials or older registries. “But when you add in the platform’s impact, you come closer to the benefit you see from controlled trials,” says Rosier.
Resource challenges and alerts
The platform involved comes into play by helping focus the massive amount of data from remote monitoring. “This is a resource issue. You have a lot of data coming in and a lot of alerts, and out of these alerts, maybe 20% are of interest,” says Rosier. “That 20% you want to see provides the value for the whole population.”
It’s like digging for gold, he says. “You need to find the signal, but it’s hidden in the noise,” says Rosier. “You have to refine it. There are a couple of ways to do that: You may have an army of clinicians looking at it, or you outsource that task, but you don’t know the quality you’re getting. Or you use AI.”
Ideally, Rosier says, the solution would leverage both a team of humans as well as AI. And these need to live in the same place to be efficient. “People underestimate the amount of time spent every day just clicking between platforms,” says Rosier.
Interoperability is step one, and automation is step two. “[After] automation, then the humans step in,” says Rosier. There simply isn’t the staffing or budget for humans to go through every data point collected through remote monitoring, but automation can distill the incoming data to make it more usable to the human expertise waiting for it.
The timing for this discussion is right, as well: The Heart Rhythm Society has recently published guidelines for remote monitoring clinic staffing, appropriate clinic workflows, patient education, and alert management. “This is the first time we’ve seen guidelines from all societies for heart monitoring, and they mention that third-party platforms are useful for providing quality assurance and resource optimizations for this activity,” says Rosier.
What is the AI looking for?
If that first pass at data is being automated, how can AI or other tools help surface important information? Rosier offers an example.
“Say one of my patients in France experiences atrial fibrillation [A-fib]. This puts the patient at a high risk of stroke. A known remedy for this is anticoagulants: blood thinners,” he says. “We need to know if the patient has A-fib. It’s about discovering if your patient is at risk. In this typical case, A-fib is detected by the device, but then there needs to be a review that it’s true arrhythmia. That’s a human task, to read the signal.”
The platform uses an FDA-approved algorithm to decrease false positives in detection. “It becomes a numbers game. We keep all the signal but decrease the burden on practitioners,” says Rosier.
There are only so many hours in the day, and thus only so much patient data human operators can review. AI can spot and escalate higher-risk patients for human review faster. “Healthcare is not a free resource game, and because society is paying for healthcare, we have a limit to that resource,” says Rosier. “And in the U.S., spendings are more than anywhere else for healthcare.”
It’s always possible to put more money into healthcare, but better monitoring and data analysis can improve the usage of existing resources to provide better outcomes for patients. “It’s important to note that AI in healthcare is not taking jobs away from people. We always need those people in healthcare to function, to do the things an AI can’t do. To talk to the patient,” says Rosier.
But if AI can take the grinding work out of surfacing data points, it can enable human practitioners to practice at the top of their license more often. “We often hear that physicians are reluctant to use new technology, but they’re actually very open to new innovations. What they want are things that work very well,” says Rosier.
Healthcare is swimming in data, Rosier says, and how it handles that data greatly impacts the care it can provide. “Compared to other businesses, in healthcare we don’t have a prewritten knowledge of what we need to do. We are building that knowledge as we move forward. Having the ability to use data to build knowledge enables us to understand what needs to be done” for the patient, says Rosier. “We build out that knowledge not just in the form of human knowledge, but also in the form of AI algorithms based on what we learn. The greatest danger in healthcare is status quo.”
Rosier notes that there’s a lot of talk about AI now—with rational worries—but he says that much like past technological innovations, it will be a game changer in ways that aren’t at the top of the conversation now. “I believe the impact will be much higher than what people think,” he says. “AI is another way of creating new sources of knowledge useful for practice.”
Matt Phillion is a freelance writer covering healthcare, cybersecurity, and more. He can be reached at matthew.phillion@gmail.com.