The Fantastical World of AI in Imaging
Here at KLAS, we talk to healthcare providers all the time. We ask for specific feedback on the solutions and services they use. Then we try to give back by sharing data on our website and offering reports. We also often ask them what kind of data they are looking for and which areas they are investing in. One of the most common questions we get is, “What do you have on AI?”
AI is a hot topic in the minds of providers. The challenge is that the market space is still early and adoption is still low, so there is a lot of confusion about what AI even is. KLAS has published a few early reports on the AI space and has worked to narrow down that definition.
KLAS’ definition* of AI is twofold: machine learning and natural language processing. Machine learning tools are like students taking an independent study class; they study computer algorithms and statistical models to learn how to do new tasks without direct instruction. Natural language processing is like a U.N. ambassador from the land of computers that has traveled to the land of humans. This technology understands, processes, and even analyzes natural language in both speech and text form.
We elaborated on the general AI space in another report. But a key piece of the hype is AI in imaging. Providers are looking for algorithms that will help them with diagnoses, workflow, task prioritization, and other innovative ways to make providers more productive.
In 2018, we published our first report on AI in imaging. After some time passed, we went back to the same topic to discover how the market had progressed, creating our recently published report, Artificial Intelligence in Imaging 2020.
Perception, Not Performance
One thing that makes these reports interesting is that they aren’t about performance; they are about perception.
Since AI in imaging is such an early market, it hasn’t really solidified yet. The interest is high, but vendors are still working on meeting that demand, and many providers are acting as test sites to implement new products. There is a lot of potential but very little solid reality.
We looked at how well providers believe they are doing at implementing AI into their organization and who they believe to be the top-of-mind vendors in this space. We will also have specific case studies by vendor to follow.
We expected that providers would report increased success compared to the previous report. What we didn’t expect was such a dramatic increase.
The fact of the matter is that very few provider organizations are fully adopting AI products in their facilities even if live or piloting. There is a crazy level of interest, but only a small number has been able to turn that interest into concrete, practical use cases. However, providers seem to be much closer to implementing AI and much more confident in their abilities to achieve their AI goals than previously.
Mindshare, Not Market Share
The opportunities for mindshare in AI are wide as the sky, but the gap is narrowing. More vendors were mentioned in the 2018 report than in the 2020 report, but most vendors were mentioned only once or twice. Now, a smaller number of vendors are being mentioned but a lot more frequently. That is what we tend to see when an innovative, theoretical market becomes more grounded in reality. Niche vendors are getting scooped up by bigger companies with enterprise platforms and app stores. The field is still large, and there is a wide variety of players.
In our 2018 report, IBM Watson Health was far ahead of any other vendor in mindshare. That gap has narrowed now—not because IBM Watson Health has done anything differently, but because other vendors like Zebra Medical Vision, Aidoc, and others are being thought of more frequently.
Vendors with a large install base of other imaging or voice solutions like GE, Agfa, IBM Watson, and Nuance are seen to be in a stronger position because of the large amounts of data, relationships and capital they have access to.
Looking Ahead
How is KLAS planning to expound upon this data? We want to compare market perception with reality by exploring live use cases. We want providers to learn from those that have been implementing AI imaging products at their facilities. We have already validated providers who are utilizing Zebra Medical Vision, Aidoc, IBM Watson and others. As we work closely with providers, we are looking forward to adding some case studies to our growing library of research on imaging AI.
*Footnote: We didn’t use the definition in our research with providers for Imaging AI 2020; we did place it in our report to provide clarity to readers and direction for future research.
Photo credit: Adobe Stock, vectorfusionart