Healthcare AI is scaling faster than ever, but it’s being held back by legacy systems – think clunky software and fragmented data – that were never designed to support AI in the first place.
AI is being embedded in healthcare at an incredible pace. In 2022, the global healthcare AI market grew by 12%, reaching a value of over $5 billion. But despite this momentum, researchers and clinicians are starting to sound the alarm about the limitations of current approaches.
“The problem is that most of these AI systems are still stuck in the old way of thinking,” says Dr. Ron Gutman, a physician and epidemiologist who founded the digital health company Evra Health. “They’re relying on smarter algorithms and better data, but they’re not actually fixing the underlying issues that are preventing us from getting real value out of AI.”
So what’s the problem? For one thing, many legacy healthcare systems are incredibly inflexible. They’re built around rigid data structures that can’t easily adapt to the changing needs of AI. And when AI is forced to work within these constraints, the results are often subpar.
“Imagine trying to build a house with a bunch of mismatched LEGO bricks,” says Dr. Gutman. “It might look okay at first, but it’s going to be unstable and difficult to work with. That’s what we’re seeing in healthcare – AI that’s being shoehorned into systems that weren’t designed for it.”
Integration headaches
The integration of AI systems is another major challenge. Right now, many AI solutions in healthcare are standalone products that don’t talk to each other. This creates a bunch of headaches for clinicians, who have to navigate a complex web of different systems just to get a basic workflow going.
Meaningful clinical impact
Finally, there’s the issue of meaningful clinical impact. Despite all the hype around AI in healthcare, many solutions are still struggling to deliver real-world results. And Dr. Gutman says that’s because we’re not focusing on the right things – we’re prioritizing the development of fancy algorithms and data sets over the actual clinical needs of patients.
What this means
So what does all this mean for the future of healthcare AI? For one thing, it means that we need a new approach – one that prioritizes flexibility, integration, and clinical impact above all else. That might involve building new systems from scratch, or it might mean rethinking how we use AI within our existing infrastructure. But whatever the solution, it’s clear that the status quo won’t cut it.
