Play

Featuring everything related to medical imaging, the Radiologic Society of North America (RSNA) is one of the country’s largest medical professional meetings and trade shows. I’ve attended the event for almost 3 decades and watched the evolution of many developments that are applicable to healthcare more broadly. Here are my takeaways from this year’s meeting.

 

First, the meeting continues to shrink.

 

It used to be that even in the massive Chicago convention center, you could barely move. Now, there are fewer attendees and definitely fewer vendors. This change is, in part, due to the growth of on-line education, but more importantly the declining role of physicians as the decision makers. Purchasing decisions are now made in bulk by hospital systems. New features—the latest and greatest—are balanced against price.

 

Second, hardware is so yesterday. Apps—more specifically AI apps—are in, but nobody knows what that means.

 

Vendors still showed off “big iron” (CT scanners, MRI units and angio suites), each costing over a million dollars. However, the market is largely commoditized with little in the way of innovation for the last two decades. With the national focus on value-based care, everyone is looking for ways to do more with less. Artificial Intelligence or machine learning promises to displace or improve physician performance. New players such as Google, IBM Watson, and Nvidia were all present.

 

There are literally thousands of AI apps being built by mom-and-pop developers all over the world. Each app offers an opportunity to help radiologists make better diagnoses. However, several challenges remain.

 

For one thing, no one has figured out a business model. There is a sense that physician practices or hospitals will pay for apps, but as yet they have no way to offset the expense. Hospitals may be required to pay based on quality metrics, but this is also ill defined. 

 

What’s more, to get adoption, apps need to be part of the clinical workflow. The focus needs to be on integration, building a marketplace managing the administrative overhead (privacy and security reviews) found in every healthcare system. IBM, if you’re listening, your booth seemed focused on apps at the expense of PACS. Please don’t sacrifice enterprise imaging, the foundation for clinical care, for the new, new thing. This approach is akin to eating dessert at the expense of the meal. The real opportunity is one step further: leveraging AI for both image analysis and related clinical information from the EMR and previous radiology reports. This, unfortunately, is years away.

 

Third, the corporatization of radiology (and medicine) is here to stay. 

 

Richard Duszak, head of radiology’s policy institute, has noted how smaller practices face significant head winds. Insurance contracting is partly to blame, but federal reporting issues (what I’ve called “death by data”) is also responsible. Large equity-backed organizations like Envision, Medimax and RadPartners are buying smaller groups. As part of this model radiologists move from being self-employed to being employees; as noted on a previous podcast, this may adversely affect medical societies, including the American College of Radiology and RSNA.

 

What do I expect to see at RSNA in 10 years? GE, Siemens and Philips, the traditional vendors financially supporting RSNA, will be less visible or have reinvented themselves to align with value interpretation rather than simply image production. The number of radiology groups in the US will have decreased, and those that remain will be tied to quality metrics. Finally, technology (AI) will be better defined, hopefully supporting the radiologist’s role as part of the clinical team.

 

My hope is that somehow we get “back to the future,” more like the time of my father’s early practice where, with the help of integrated clinical information, radiologists will again be the physician’s physician. I hope to be in Chicago for that meeting.