Today wraps up the 104th annual Radiological Society of North America (RSNA – www.RSNA.org) meeting.Mother Nature made it a challenging place to get to early in the week, but from all accounts, attendance was on par with the past few years.From an imaging informatics perspective, this year saw a number of things that point to a resurgence in imaging. It also presented some disappointment with respect to how the imaging vendors are dealing with a changing healthcare environment.Artificial Intelligence – the obviousLet’s begin with the 600-pound gorilla in the room, and that would be Artificial Intelligence (AI). By all accounts, if you were to sum up this year’s meeting, AI everywhere would be how one would describe it! AI has been a topic of discussion for several years now, initially driven by IBM’s Watson Health initiative. In prior years, there was considerable talk about how AI was going to revolutionize Radiology, and potentially replace the radiologist. This year, the emphasis seemed to really shift from the “pie-in-the-sky” discussion to real-world, commercially available solutions.
A key development conundrum has been how to commercialize AI. Academic centers represent a first line of research into AI applications, while “boutique” companies have struggled with how to get developments to market. Large imaging informatics companies have likewise wrestled with how to approach bringing AI applications to market. The solution prevalent this year seems to be for both large and small companies to offer a “platform” for the implementation of AI. By supporting such capabilities as software development toolkits (SDK’s), vendors are providing a means for commercialization of academic and third-party applications without themselves reinventing the wheel.The AI “store” borrows from the way smart-phone applications have evolved by providing the infrastructure for the validation and distribution of AI applications. What is not yet clear is the liability of providing access to other entity’s applications. Is the Store vendor responsible for the application, or the developer? Who files for and secures FDA approval? Given that the objective is for these external applications to interoperate with the vendor’s imaging informatics system, there is some development risk on the part of the distributing company, and potentially a shared liability as well. Only time will tell how effective this strategy is.Depending on who you ask, AI primarily is perceived as clinical tools to improve the radiologist’s interpretation efficiency, not as a replacement to the radiologist from a clinical perspective. Conversely, there were a number of applications that make use of AI technology to enhance the way information is handled and presented, and the way it impacts the decision process. Much of this revolves around the way information is collected and made available to the clinician, such as retrieving relevant lab and other study information. One interesting example might be Siemens Healthineers’ Proactive Follow-up application (https://usa.healthcare.siemens.com/healthineers-population-health-management/value-based-care/proactive-follow-up-for-incidental-findings).It uses natural language processing to identify incidents of follow-up, such as “repeat CT exam in six months.”Incidents requiring follow-up are summarized in a “dashboard” presentation to enhance the ability of imaging services to coordinate with the necessary clinical services to ensure that the follow-up recommendation is followed through.While not as “sexy” as an AI image processing algorithm, it may have just as much if not more impact on imaging services’ efficiency.AI will influence imaging in another way by fostering greater use of the cloud. To maximize availability and accessibility, the cloud appears to be the major means for the deployment of AI applications. Some vendors are also increasingly moving to the cloud for their entire enterprise imaging informatics applications. Such non-traditional players as Intel and Google are becoming a greater factor in terms of how imaging is secured and managed, and AI appears to be an influencing factor. Clinical Decision Support – the not so obviousWhile major emphasis was on AI, less emphasis seems to have been given to Clinical Decision Support, and the associated mandates. The Protecting Access to Medicare Act of 2014 (PAMA) originally directed CMS (Centers for Medicare and Medicaid Services) to require Appropriate Use Criteria (AUC) consultation for Advanced Diagnostic Imaging procedures beginning Jan 1, 2017. The mandate has now been delayed to January 1, 2020, which isn’t that far away!Imaging companies correctly point out that clinical decision support will be more a function of the electronic health record (EHR) system, and they don’t seem to be particularly concerned with how it will impact imaging applications, with a few notable exceptions.Change Healthcare (https://www.changehealthcare.com/) has been reformulated over the past few years from a “back office” services company to one encompassing imaging through the acquisition of McKesson’s imaging business.More recently, Change acquired National Decision Support Company (http://nationaldecisionsupport.com/) to address the PAMA mandates by means of synergy between its product lines.Similarly, Siemens Healthineers acquired Medicalis, which was also focused on clinical decision support tools. Collectively, these two vendors seem most aggressive in addressing the intersection of imaging services and the changing landscape of healthcare management.Value-Based Care – another not so obviousHealthcare providers are moving away from fee-for-service models to value-based care models of healthcare delivery. These changes will ultimately impact imaging services, yet there appeared to be little direct emphasis amongst exhibitors. Part of this conundrum may be the perception that much of the informatics needed to address value-based care will be encompassed within the EHR. On the other hand, imaging vendors seem to be more focused on the “mechanics” as opposed to the topic of value-based care. For example, analytics tools and intelligent worklists are mechanisms that will help enable radiology to support value-based care, but they are not necessarily emphasized as such.Consolidation and New PlayersThe industry continues to be a study in competitive dynamics, in that certain segments demonstrate further consolidation, while other segments continue to expand.The area of workflow orchestration has seen a transition from “incubator” companies such as Clario, Primordial, and Medicalis to complete absorption by large imaging vendors.Siemens Healthineers previously acquired Medicalis, and Nuance acquired Primordial.The surprise announcement at this year’s meeting was the acquisition of Clario by Intelerad (https://www.intelerad.com/en/press-releases/intelerad-medical-systems-acquires-clario-medical/).This now means all three of the key workflow orchestration vendors are part of larger imaging informatics organizations, and can leverage those capabilities as part of their offerings. On the other side of the spectrum was the dramatic introduction of United Imaging Healthcare (https://usa.united-imaging.com/united-imaging-healthcare-makes-u-s-market-debut-at-rsna-2/) into the U.S. market.United made their entry with one of the larger exhibits, and a dramatic first-day unveiling.While operating in other world markets prior to this year, United has made a large investment by establishing a U.S. presence.For a number of years, the imaging industry has lived in the shadow of the EHR, as providers scrambled to address government mandates for electronic health records. Now that much of that infrastructure is in place, it appears that imaging informatics may be well-positioned to capitalize on further investment to support the EHR. AI appears to be the first recipient of that emphasis. From my vantage point, there will need to be a further shift to emphasize applications and solutions that support consolidation and value-based care trends. It will be intriguing to see if these areas receive more emphasis at RSNA 2019!
