Elevating Radiology’s Influence Across the Health System with AI

In April of 2021, Aidoc and Radiology Partners officially formed a partnership to determine how artificial intelligence (AI) could elevate the radiologist's influence across the healthcare system. Through that process, both companies learned a few lessons along the way.

In an AI Showcase presentation titled “Elevating Radiology’s Influence Across the Health System with AI: Aidoc,” Nina Kottler, MD, MS, and Associate CMO for Clinical Artificial Intelligence at Radiology Partners, outlined those three lessons in choosing an AI product and partner.

First, she advised that radiology practices find a partner, not a product. She noted that AI is still an immature area, where there are few best practices or standards of care. This makes it critical to base business decisions in partnerships, not just technology.

“You can't partner with an AI algorithm, so you need a partner in this process,” said Dr. Kottler.

“When you have to create standards as you go, it’s far better to have a partner to do that with than an algorithm. Because an algorithm can't help you, but a partner absolutely can.”

She explained that partners should be able to help radiology practices solve problems, work collaboratively and be excited to learn from your clinical expertise, as you learn from their technological expertise.

“If you're a radiology practice, you have a huge amount of clinical expertise. The vendors have a huge amount of technical expertise and expertise in the data science domain. When you combine them, they're better together than alone. So make sure you're working with someone that cares about that,” said Dr. Kottler.

An AI partner should also share your mission and values, and they should recognize the radiologist as the central position in healthcare. “Radiologists are centrally located within healthcare, so make sure that you're working with a vendor that values that position,” she explained.

Second, use technology to bring specialties together – without bypassing the radiologist.

“AI is not always right. And radiologists are not always right. But when they work together, they do better than either alone. So you want to make sure that the radiologist is still providing care and confirming findings and helping determine treatments,” she said.

Third, invest in radiologists’ AI education now – not later. “Take the time to educate your radiologists and learn about AI because we need to take the driver's seat and help AI create the best outcomes for patients, clinicians and hospitals. Take that time to learn more – and go from the early learner or early adopter to become the early expert,” she advised.

She noted that Radiology Partners are doing this at scale with Aidoc in creating a cloud native solution that allows them to move data in real-time, connect with an AI solution, and then reconnect with clinical applications that contains Aidoc’s algorithms.

“We have 15 million images and reports available to this algorithm. We send them into our cloud native platform, and Aidoc evaluates the image and report to determine what the radiologist says. We can then compare the two. If there is a discrepancy between the two, then we can pass that into our AI-enabled peer learning system from which we can educate our radiologists. Because you need to educate people in order for the systems to work better together,” she said.