This Insight is part of your subscription to Signify Premium Insights – Medical Imaging. This content is only available to individuals with an active account for this paid-for service and is the copyright of Signify Research. Content cannot be shared or distributed to non-subscribers or other third parties without express written consent from Signify Research. To view other recent Premium Insights that are part of the service please click here.
Co-written by Dr. Sanjay Parekh
At RSNA 2021, IBM Watson Health revealed a new cloud-based AI orchestration platform, which the vendor says will allow health networks to deploy and use medical AI applications more easily and efficiently. The new platform is vendor neutral and, IBM believes that along with its existent marketplace, will help providers solve the ‘last mile’ challenges of selecting, deploying and integrating imaging tools from multiple AI developers into their clinical workflows.
The Signify View
Signify Research’s Machine Learning in Medical Imaging report has, for several years, highlighted that one of the key barriers to greater AI adoption is the current lack of its integration into clinical workflows. The advantages of a key AI solution could be myriad and profound, but that will count for naught if providers are faced with challenges around integrating these tools and doctors are unable to utilise the tools easily and effectively in their everyday clinical practice.
The impact of this lack of integration could be particularly acute in imaging, where there is both a growing volume of diagnostic imaging, and distinct shortage of radiologists, often leading to intense time pressures. Adopting AI into the clinical workflow to address these challenges may improve productivity, but this will be negated if solutions require physicians to leave and enter distinct software packages, wait for tools to run, and make extra clicks. AI orchestration platforms aim to overcome these problems, by, for example, curating AI solutions that provide greater value to the radiologist, enabling them to focus more on patient care.
IBM Watson Health’s new AI Orchestrator platform aims to minimise these inconveniences, with the platform making it easier for providers to leverage the advantages of a range of AI tools, without deploying and integrating them individually. Under IBM’s AI Orchestrator, solutions from developers which have partnered with IBM such as Cortechs.ai and Behold.ai (under development), will be seamlessly available to users of IBM’s Orchestrator, regardless of their PACS or other informatics systems. However, curiously, there is still a disparate approach from the company, as it has not yet incorporated any of its AI marketplace partner applications onto its AI orchestration platform. Whether IBM chooses to do so remains to be seen, at this instant it seems a missed opportunity in terms of swiftly expanding the value of its AI orchestration platform at its launch.
One of the notable things about IBM’s new solution is that it is entirely cloud-native. While increasing numbers of vendors have developed cloud native AI platforms, and, as detailed in Signify Research’s Cloud Adoption in Imaging IT report, it is one of the clear directions in which the medical imaging IT market is heading, IBM’s relative strength in the broader cloud infrastructure could be an advantage. Furthermore, the AI orchestrator platform was announced in sync with a new broader cloud-native workflow platform, Imaging Workflow Orchestrator, in which IBM has brought to market what it believes will the new industry standard in diagnostic workflow environments.
This solution is, in some ways, more important than the vendor’s AI platform, given IBM’s additional exposure to the PACS and VNA markets compared to the AI platform. The Workflow Orchestrator brings some significant features to IBM Watson Health’s portfolio, namely integration of advanced enterprise worklist (including AI triage results), diagnostic viewer and pertinent EMR-based patient data (Watson Patient Synopsis) into a singular user interface. This is, in part, achievable because the new platform is heavily tied to the IBM VNA. Watson Patient Synopsis, utilises AI to identify the most relevant data from a patient’s record in an EMR system and makes it readily available to a radiologist, addressing an increasing focus in diagnosis in bringing relevant patient data to front line diagnosticians in a digestible format. The integration of these functionalities, and the fact that it is cloud native, gives IBM an advantage over some similar enterprise radiology products from the other larger informatics vendors, which are less progressed in terms of their cloud native strategy; Workflow Orchestrator’s seamless integration with the AI Orchestrator could also give the vendor a further competitive edge.
IBM also hopes that its cloud expertise within the broader company, particularly on the back-end through its acquisition of Red Hat in 2018, should help sway customers. Some providers might be wary of entrusting their precious medical data to public cloud from the likes of Google, or Amazon, but being able to gain cloud capability from the same vendor that facilitates their AI toolset and broader imaging IT platform could be enough to convince them. As such IBM Watson Health will hope that the new AI Orchestrator, Workflow Orchestrator and its cloud capability will drive business to the vendor and bring with it opportunities to sell additional products. However, IBM is also not wed to its own cloud services as the new AI Orchestrator can be deployed on any cloud.
Many providers, however, are not yet ready for full cloud deployments. Some have invested heavily into their own on-site data centres or have preference for keeping their data on site for other reasons. There are others still which aim to take advantages of the benefits of cloud-based deployments, but do not wish to entirely give up on premise solutions and so are looking to utilise hybrid deployments, either as a temporary measure on the way to complete cloud nativity, or as a preferred permanent solution. IBM’s new AI Orchestrator can facilitate these hybrid cloud deployments, opening up its use to a far greater range of providers and allowing for more flexible deployments. However, whether the need to accommodate a range of different deployment architectures will harm its functionality, remains to be seen, but it is critical the availability and performance of its solutions aren’t offset by its need to appeal to a very diverse range of providers. This could simply lead to a poor experience for a greater range of people.
There are also other challenges set to weigh on IBM’s strategy. One of the keenest is the increasing levels of competition from both other vendors and other solutions. There are now several platform solutions available. Some offer advantages that are unique, or at least uncommon. Aidoc’s, for example, focuses on a means to deploy its growing portfolio of native applications, supplemented by AI solutions from third-party vendors. Blackford Analysis, on the other hand, has curated a much broader ecosystem of applications from third-party vendors, with the ability for providers to select the AI solutions that work best on their populations and case mix.
Other platforms created by larger imaging IT vendors, such as GE Healthcare and Sectra, are potentially better integrated into a broader imaging IT workflow but focus on partnering with AI vendors for their content. On the other hand, vendors such as Siemens Healthineers have chosen to develop most of their AI tools in house, carefully creating and curating only a choice selection of native and close-partner integrations, ensuring they are fully integrated into the vendor’s wider imaging IT solutions.
IBM will have to demonstrate real additional value and a unique proposition if providers are going to select its AI Orchestrator over these tools from competitors. This could come from the platform itself, or from the range and quality of AI tools integrated into the platform. Or, as is more likely the case, from leveraging its Patient Synopsis and Clinical Review tools that integrate upstream and downstream of the clinical workflow. Watson Patient Synopsis, utilises AI to identify the most relevant data from a patient’s record in an EMR system and makes it readily available to a radiologist. Clinical Review is a retrospective analysis tool that compares the radiologist’s report (including image analysis) to determine potential areas for reconciliation.
A Persuasive Proposition
Whether it can achieve this and therefore secure commercial success remains to be seen. This question is particularly pertinent given the year’s earlier rumours of a sale of IBM Watson Health. A previous Signify Premium Insight commenting on the rumours suggested that one of the motivations for the sale was the broader company’s increasing focus on cloud. Offering customers a desirable cloud-native platform could bolster sales in the company’s cloud business, in much the same way as Microsoft’s acquisition of Nuance is at least partially driven by the big tech firm’s ambition to use healthcare to bolster its own cloud business.
Regardless, IBM Watson Health’s platforms are unlikely to materially change the fortunes of the business unit, and ultimately a sale is unlikely to prevent its software from being run from IBM’s cloud. But, the fact that IBM is still investing in its R&D suggests that the businesses fate is not yet sealed.
As it stands, the vendor has a mid-single-digit share of the North American medical imaging IT market, this share could render it difficult for IBM to have a sizeable market impact by only focusing on its own Imaging IT customers. Instead, the new AI orchestrator’s vendor neutrality should be emphasised, and IBM should highlight the platform’s ability to enhance other vendors’ packages and provide an effective software infrastructure to providers looking to add AI capability. Moreover, the vendor’s cloud focus in its foray into AI platform deployment could also prove beneficial and could form an integral part of a future cloud-based enterprise imaging deployment.
For the time being though, IBM has brought together a lot of existing capability into a single, cohesive, cloud-native package. Many providers will pay little mind to this development, and duly wait for comparable capability from their current imaging IT vendor. The relatively few integrated AI applications may also leave some potential customers waiting a little longer to see how quickly the firm can scale up to a more comprehensive AI partner ecosystem, especially with no version combining the marketplace and orchestrator yet available. For others, however, IBM’s AI Orchestrator brings together its existing upstream and downstream AI capabilities within a single, flexible platform, promising streamlined participation in the growing trends of cloud and AI. This may just be enough to help providers forgive the overpromising of the past on AI and convince them of a potentially bright future second time round.
About Signify Premium Insights
This Insight is part of your subscription to Signify Premium Insights – Medical Imaging. This content is only available to individuals with an active account for this paid-for service and is the copyright of Signify Research. Content cannot be shared or distributed to non-subscribers or other third parties without express written consent from Signify Research. To view other recent Premium Insights that are part of the service please click here