Data mining helps hospitals identify new patients in need of TAVR, other structural heart procedures
When the American College of Cardiology and American Heart Association collaborated on new valvular heart disease (VHD) guidelines in 2020, the two groups highlighted the importance of interventional treatment options such as transcatheter aortic valve replacement (TAVR). However, an estimated one-third of patients with severe aortic stenosis (AS) in the U.S. still lack a guideline-recommended treatment plan, dramatically lowering their chances of survival.1
Charnai Sherry, PA-C, who specializes in cardiovascular disease, says it is absolutely crucial to treat symptomatic severe AS patients as early as possible.
“When you find a patient with symptomatic severe AS, it’s important to treat them right away,” she says. “As time goes by, their risk of heart failure or another serious complication only increases. Before you know it, it’s too late and you have to do an emergency procedure instead of a standard interventional procedure.”
Echo mining increases patient volume
Sherry is the valve clinic coordinator with the structural heart program at Aurora St. Luke’s Medical Center in Milwaukee, part of the Advocate Aurora Health system.
Aurora St. Luke’s is known for its massive TAVR program—performing 500 to 600 procedures per year—as well as its experience with mitral and tricuspid procedures. The structural heart program also regularly participates in key clinical trials.
One of the many reasons for her team’s success, Sherry says, is its continued commitment to data mining. Aurora St. Luke’s has been mining echocardiography data for years using special “Best Practice Advisory” alerts built into her team’s electronic medical record (EMR). The tool was built in house by the hospital’s IT department.
“If there are patients out there who need us, we want to find them. Mining echo results helps put us in the driver’s seat so we can accomplish that goal and make an impact.”
Charnai Sherry, PA-C, valve clinic coordinator with the structural heart program at Aurora St. Luke’s Medical Center in Milwaukee
Here’s how it works: Once echo results are imported into the EMR, any exams that meet the prespecified parameters are automatically flagged by the system. A clinician is then advised to work toward getting that patient into a structural heart or cardiovascular surgery specialist for further evaluation, which can be done with a single click.
“Data mining has been incredibly important for our entire structural heart program,” Sherry says. “It identifies untreated patients and helps us bring them in for an evaluation right away.”
Referrals to the structural heart program at Aurora St. Luke’s increased by 20% when they first started mining echo results several years ago. Fortunately, Sherry says, clinicians are able to filter and sort through those patients much quicker than ever before.
“We have a lot of flexibility with what we can do with the data,” she says. “We can confirm a patient has severe AS, explore their medical history, see who referred them—if we have any questions at all, we know how to quickly find answers. It definitely makes our jobs much easier and makes us much more efficient.”
As the TAVR program at Aurora St. Luke’s continues to thrive, Sherry says she has also noticed significant growth in interventional treatments involving the mitral, tricuspid, and pulmonic valves. Data mining, she thinks, will play an important role as her team works to find patients in need of those services as well.
“If there are patients out there who need us, we want to find them,” she says. “Mining echo results helps put us in the driver’s seat so we can accomplish that goal and make an impact.”
AI-powered echo mining: A new solution to an old problem emerges
Kristin Pasquarello, MPAS, PA-C, administrative director of The John Brancaccio Heart Valve Center at St. Francis Hospital and Heart Center in Roslyn, N.Y., has also experienced the benefits of data mining. Pasquarello and her colleagues at St. Francis regularly complete more than 500 TAVR procedures per year, but they are always looking for new ways to identify patients with untreated severe AS.
“We can significantly improve quality of life with treatment,” Pasquarello says. “It’s so important to see these patients, so we can determine what it is they need.”
The structural heart program at St. Francis is busier than most hospitals—they perform and interpret three times more echo exams than the average hospital—so its leaders hoped to go beyond customizing their EMR and bring in a state-of-the-art data mining solution that could handle their high volumes without missing a beat.
With this goal in mind, as well as a commitment to exploring advanced technologies whenever possible, St. Francis decided to try a data mining solution developed by Mpirik for a four-month test run. Mpirik’s Cardiac Intelligence platform uses advanced artificial intelligence (AI) algorithms and natural language processing (NLP) to scan patient data and identify patients with severe AS who may not be receiving the care they need. The user sets up the parameters—specific echocardiography findings, in the case of St. Francis—and then the platform performs a thorough search of all available patient data, identifying anyone with severe AS who is not already receiving guideline-recommended care. Once a patient is identified, their primary care physicians are told that they may require screening and care.
The platform’s AI and NLP capabilities were especially appealing to the team at St. Francis. If they could identify untreated or undertreated patients with Mpirik’s advanced algorithms, they knew it would be a significant step forward for patient care. However, there was still a certain degree of skepticism at first at St. Francis—not because they doubted the technology, but because they were so confident in their own abilities.
“Our TAVR volume was already one of the highest in the area,” Pasquarello says. “I didn’t see how that many patients could be falling through the cracks.”
"Mining echo data helps us spread the word about what our heart center can do."
Kristin Pasquarello, MPAS, PA-C, administrative director of The John Brancaccio Heart Valve Center at St. Francis Hospital and Heart Center in Roslyn, N.Y.
There also were some concerns about how other providers may react to a data mining tool being used on their patients. Would physicians think St. Francis was trying to steal their patients away? Would radiologists speak out about having their echo reads poured over by AI and NLP?
But it quickly became clear—to Pasquarello and everyone else who saw the solution in action—that data mining was a success. St. Francis started identifying yet-to-be treated severe AS patients right away, helping get those patients on the road to recovery. Meanwhile, referring physicians learned to trust data mining’s discoveries, and radiologists saw it as a helpful quality initiative.
Pasquarello says the data mining tool is “100% customizable,” allowing the user to filter out anything they don’t need and focus on the exact targets of their searches. It also learns from its users as time goes on, remembering the details of previous searches to improve the customization process.
“In a way, it mines a lot deeper into the data than we would on our own,” Pasquarello says.
And while TAVR patients were the primary target of its early efforts, St. Francis is now able to use the technology platform to find more patients who may require other, less common structural heart procedures.
“Awareness about TAVR is pretty good right now, but that isn’t the case yet with some of the newer procedures that are out there,” Pasquarello says. “That makes it even more important to identify patients who need our help. A lot of primary care physicians and even general cardiologists just don’t know everything going on in, say, the mitral space or tricuspid space. Mining echo data helps us spread the word about what our heart center can do.”
After the four-month trial, the Mpirik offering had reviewed more than 5,000 echocardiography exams, identified 31 new patients with severe AS, and directly led to eight new patients undergoing TAVR at St. Francis. An additional seven new patients underwent mitral valve interventions, and two more new patients underwent balloon valvuloplasty procedures.
The clinical and administrative teams both chose to invest in Mpirik long-term.
“Mpirik allowed us to identify new patients who otherwise would not have been identified. Through our own analysis, we found that the cost of Mpirik was paid for by performing four extra TAVR procedures per year,” Pasquarello adds. “Knowing how easy it would be to hit that number, everyone involved agreed it was a smart idea to invest in the platform and move forward.”
More information on Mpirik is available here.
Disclaimer: Operational, clinical, and financial impact calculations were provided by St. Francis Hospital. Results may vary and depend on site implementation of recommendations.
1Mpirik data on file
Michael has more than 18 years of experience as a professional writer and editor. He has written at length about cardiology, radiology, artificial intelligence and other key healthcare topics.