AI is reshaping medical device data triage by finding the signal in the noise

Whether you’re treating a panel of patients or just sorting through the spam in your morning inbox, top-notch triage skills have become essential for dealing with the constant data overload inherent in modern life. 

Nowhere is this more apparent than in the world of electrophysiology, which is solely focused translating the fundamental electrical pulses of living tissue into human-readable signals – and then interpreting those signals to make decisions in the context of complex clinical activity. 

Technical and clinical advances have brought a huge number of new medical devices into the world of patient care, which have, in turn, infused even more data into the diagnostic and treatment process.  

But not all data is created equal, says Nick Von Bergen, MD, a pediatric cardiac electrophysiologist at The University of Wisconsin-Madison and co-founder of Atrility Medical. 

And even the best clinicians can benefit from a helping hand to identify the data that matters most in the moment for making specific choices about care, especially when pediatric patients are involved.  

“Not all devices offer the same quality and relevance of data for every potential clinical concern, so the risk is that we get overwhelmed with low-quality or overly voluminous data and we miss important findings simply because they’re buried,” he told Digital Health Insights. “That’s the real danger: we don’t want to miss something life-threatening because it’s not visible under the pile.” 

“AI is very good at extracting signals from large volumes of data.  They can detect patterns that humans might miss, especially when signals are subtle or intermittent. If we can filter and prioritize data before it reaches the clinician, we can reduce overload while still capturing important findings.” 

The challenges of triaging device data in the current environment

In cardiac electrophysiology, clinicians monitor patients with a variety of devices that send back digital information that must be analyzed within a certain timeframe, Von Bergen explained. These devices, which include a growing number of remote monitoring tools, are essential for expanding access to care while allowing young patients to live their lives as fully as possible. 

However, expanded access means that more patients are now coming to centers that may not be expanding their workforces at the same rate, leaving clinicians to sift through enormous volumes of data for each individual.  

“Right now, a lot of the burden is just navigating systems to get to the information we need,” he said. “Even small workflow inefficiencies add up really fast to a poor experience for the provider. Fewer clicks, fewer logins, better integration…those things matter when you have a 14- to 30-page report that you have to work through manually.”   

“That’s time-consuming, especially when much of that information hasn’t changed from the last time you reviewed it.  It’s not an efficient use of time or brain power to have to pull out the few key nuggets you need from all of that background information.” 

AI tools present a promising opportunity to shorten and smooth out those workflows by synthesizing and analyzing disparate data streams, then presenting insights in a way that’s easy to interpret quickly. By using tools such as Murj, which help consolidate data from all implantable and wearable cardiac devices into a single platform, Von Bergen’s team can navigate the high volume of incoming data while reducing the burden of manual review.

“Instead of reviewing long reports, we could move toward condensed views that highlight what’s important using flags or visual cues to direct attention where it’s needed,” Von Bergen said. “If key parameters haven’t changed and/or are within normal range, that information could be summarized rather than manually reviewed each time, allowing us to focus on exceptions.” 

Getting to the next level of AI-assisted triage and decision-making

AI can help clinicians to zero in on the most important signals in the data – but only if the tools are designed to prioritize transparency and preserve the central role of human clinical decision-making, Von Bergen stressed. 

“Right now, we’re not at the point where we can fully trust AI systems to do this work completely independently. If something is potentially dangerous, the accuracy has to be extremely high—essentially at the level where multiple experts would agree. That has to guide how we design and use these systems, and we’re not there yet in terms of technical maturity.” 

“As we progress toward better and better tools, they need to be transparent enough that we can interrogate the diagnosis we’re given and look under the hood to verify how it got there.  I would always want the ability to verify what the system is doing. There’s still a layer of decision-making that requires human judgment.  Clinical care is about understanding what matters for the patient, not just what shows up in the data.”  

Looking toward the future of AI-enabled clinical care

AI-assisted data triage capabilities for medical devices will help clinicians accelerate innovation while providing quality care to more patients with fewer burdens. 

“We’re going to have the ability to wrap a digital layer around both inpatient and outpatient care with more monitoring, more data, more integration,” said Von Bergen. “I hope that as we move forward, we can get so accurate that we will begin to trust it more, while still having the ability to verify.” 

“If we can improve how data is acquired and interpreted, we can bring higher-level diagnostics to places that don’t currently have them.  The accessibility of care, the miniaturization of devices, the ability to get better data in more places…that’s going to make a big difference for patients,” he concluded. “We don’t even know everything that’s coming down the pipeline yet. There’s going to be a lot of change, and it’s happening very quickly.  I think that’s one of the more exciting parts about being in medicine right now.” 

Jennifer Bresnick is a journalist and freelance content creator with a decade of experience in the health IT industry.  Her work has focused on leveraging innovative technology tools to create value, improve health equity, and achieve the promises of the learning health system.  She can be reached at [email protected].