Radiology Partners, the largest physician-owned radiology practice in the United States, has formed a strategic alliance with artificial intelligence solutions vendor Aidoc to accelerate the adoption of AI as the standard of care in radiology.
Headquartered in Southern California, Radiology Partners has approximately 2,800 radiologists providing services to more than 1,750 hospitals, clinics and imaging centers across 33 states.
In an interview with Healthcare Innovation, Nina Kottler, M.D., Radiology Partners’ associate chief medical officer for clinical artificial intelligence, explained where AI is having the most impact in radiological practice today.
She noted that AI can either use its computer power to look at an image and derive information from that or it can look at words like a report or information in the EHR and through natural language processing derive information from that.
The computer vision aspect allows it to look at an image to find some kind of pathology in order to do a few things: alert the radiologist or to move that study higher on the work list because if there is a critical finding, it's important for those patients to be seen and that exam to be opened sooner, she said. “That's very different from how we do workflow right now,” Kottler said. “Our workflow is pretty basic: we order our examinations based on when the study comes into us and when the study is due. And when that is due is just based on service level agreements and nothing related to what's in the exam. Now we have the ability for an AI system to look inside the exam and tell us there's something that's really important and alert you so that you can look at that more quickly. That's been a really big, powerful effect.”
AI also can look at the EHR and summarize information about the patient, which can provide a much more robust history than radiologists currently get, Kottler said.
Elad Walach, CEO and co-founder of Aidoc, said the partnership with Radiology Partners would “supercharge” the adoption of AI, not just among early adopters but the majority of the market. “We understood that hospitals and health systems do not want to consume a lot of point solutions. They are looking for an enterprise approach to AI. We’ve been focusing on providing a comprehensive suite of tools, beginning with acute critical findings, which is what Dr. Kottler spoke about. In this partnership with Radiology partners, we can actually bring that to an even deeper level for their health system customers because they get the complete package, including help with training and education, and help with deployment.”
As Radiology Partners has scaled up since its founding in 2012, it has been making investments in technology and clinical innovation, with the goal of significantly transforming the value that radiology adds to the healthcare field, said Rich Whitney, chairman and CEO. “We see a huge opportunity in front of us over the next 10 to 20 years, where radiology, which is probably already the most tech-enabled segment of healthcare services, has the opportunity to take that to a whole new level with the advent of AI,” he said. “Computer vision-type technologies are developing incredibly rapidly, not just in healthcare, but outside of healthcare even faster. Having the ability as a large-scale provider to make those investments that are necessary in order to be able to make what sounds good and exciting into actual reality that helps patients and the healthcare system is why we're really excited about this announcement.”
Radiology Partners looked at several AI vendors, Kottler said. “It's not just about looking at who has the most solutions or who has a good relationship with the FDA. It's about picking a partner; it's about the group and the vendor and not the actual algorithms. We've been exposed to a huge number of different AI vendors,” she explained. In 2019 and 2020 Radiology Partners ran a six-month pilot with three of Aidoc’s products and did a pretty robust evaluation of the value.
“Ultimately, if the healthcare system is going to pay for these things, they need to show value — and that's not only improved quality and patient outcomes, which is essential, but you also have to do it at a decreased cost, or an increased financial incentive for the hospital client," she said.
In their pilot they found that the radiologist plus the AI system together were more sensitive and picked up exams that would have been missed by the radiologist alone. “We found that some of those exams could have had very bad clinical outcomes,” Kottler said. “We found that the radiologists enjoyed using these tools and that they had less burnout,” she added. “We found that they also could save money for the healthcare system overall. It could decrease length of stay; it could increase patient throughput through the ER. So there was a lot more value than we expected.”
In an organization as big as Radiology Partners, there's always a challenge of rolling out something this significant. “It’s important to realize that these are clinical tools,” Kottler said, “and just because they drive value and they're super cool, and we think they're fantastic, it doesn't mean that every radiologists is actually going to use them. No matter how good a tool is, there's always a component where people are resistant to change. So you have to roll them out with a change management process. That’s true with any clinical tool, but I will say with a tool that affects patient care, you also have to make sure you're educating the radiologist, and the radiologist has to be the master of the tool.”
In the AI space, vendors on their own are going to have a very difficult time reaching any kind of economies of scale or significant penetration within the healthcare system, Whitney said. “And similarly, a radiology practice like us without any particular expertise in AI development also has a missing piece. Combining the two really solves a lot of problems that have caused AI adoption to be slower than it otherwise would be and would be barriers going forward to increase AI,” he added.
The partnership is combining Radiology Partners’ clinical expertise and large sets of real-world data with Aidoc’s talented set of AI engineers. “I think that increases the odds that we're going to tackle some really important problems and make a lot of headway.”