Getting real about AI in healthcare

If you’ve read anything about HIMSS19, then you already know artificial intelligence (AI) was the new shiny topic that dominated the conference. Initially, this troubled me because it seemed like attendees and exhibiting vendors were taking focus away from some of the very serious interoperability issues that still plague the healthcare industry and were placing them on some conceptual AI applications that could take years to materialize.

However, a session I attended entitled Reactions from the Field: AI changed my tune on the subject. In this panel discussion, thought leaders from IBM WatsonHealth, InterSystems and Hyland Healthcare shied away from the hypothetical and focused instead on realistic applications of AI in today’s healthcare landscape.

AI: The great waste-buster

The panel agreed that while the potential for AI in healthcare is exciting, there is still a long way to go until the promise of the technology is fully realized.

“The winter of AI is over,” says Anil Jain, MD, VP and chief health information officer for IBM WatsonHealth. “This was the time period where there was great despair and health systems were being overrun with data. We’re not in the summer of bliss yet when it comes to AI, but we’re beginning to see the computing capabilities, neural networks and use cases come together.”

The winter of AI is over.

To the panelists, AI has become a means for healthcare providers to deal with the deluge of data they face in many clinical and administrative areas. As such, the real value proposition of AI today is reducing waste and optimizing workflows.

“The key to AI success in today’s healthcare environment is to start by applying it to mundane issues like administrative process optimization or data consolidation and summarization,” said Jeff Fried, director of product management for InterSystems. “AI isn’t all about cancer detection. Much of its immediate value is in enabling things like patient wait time reductions and readmission prevention.”

Thinking of the technology in this manner can also help alleviate many of the concerns clinicians have about AI replacing them.

“AI isn’t about replacing doctors, it’s about making them more effective,” said Razvan Atanasiu, CTO at Hyland Healthcare. “If there are tedious tasks they can streamline or eliminate using AI — such as manual documentation and data search — then they should go for it and free more time to spend with their patients.”

Data or insights — what’s more important?

While the panel agreed on the current state of AI in healthcare, they were at odds with regard to the role data plays in the process. Both Fried and Atanasiu believe data is critically important to current and future AI processes.

“AI is the engine, but data is the fuel,” says Fried. “You need the data that feeds AI algorithms to be comprehensive and clean.”

“You can’t overlook the data,” says Atanasiu. “You build AI from the ground up, starting with the data. Also, you need to have interoperability. You need to have the APIs that provide several systems and users with the same access to the underlying data to make AI work.”

Jain, on the other hand, felt that while important, data pales in comparison to the insights AI can deliver. From his perspective as a clinician, he feels most health systems don’t need more data. They need the insights AI can provide when algorithms are applied to the data they already have.

What does the future hold?

The immediate value of AI in healthcare may not be the futuristic applications you read about in magazines and newspapers, but the panel believed that these solutions are coming. For example, the panelists envisioned scenarios where an Alexa-type, voice-activated virtual assistant could be driven by AI and applied to a variety of clinical processes. In fact, some believed that this assistant could even become smart enough to answer the question a clinician meant to ask as opposed to the one actually verbalized.

However, bringing these applications to fruition will be rife with challenges. Perhaps the most pressing hurdle will be finding the next generation of data scientists and programmers who can bring these types of solutions to life.

“The variables in AI are always changing,” says Atanasiu. “You need the next wave of technology leaders to keep up with the pace of technology, and then you have to convince them to work in healthcare instead of another field. There is an extreme demand for qualified talent right now. One that most healthcare organizations seeking to leverage AI are having difficulty addressing.”

It’s a real challenge with real consequences for all of us.

Ken Congdon has expertise in the healthcare technology industry and has been a contributor to the Hyland blog.
Ken Congdon

Ken Congdon

Ken Congdon has expertise in the healthcare technology industry and has been a contributor to the Hyland blog.

... read more about: Ken Congdon