The hype surrounding AI might have you believe the technology is rapidly transforming the healthcare industry for the better. But in reality, most health systems aren’t seeing a huge ROI from their AI investments yet.
There are three changes that must happen in order for health systems to start seeing a greater ROI when it comes to AI deployments, according to Michael Kalishman, chief venture officer at Virginia-based Sentara Health. He outlined these changes during an interview last week at the ViVE conference in Los Angeles.
AI vendors should focus more on making sure their tools actually address health systems’ biggest pain points — and less on hawking the great things their products can do.
There is a “real disconnect” between AI companies’ ability to promote the features that their products can deliver and their ability to match those features to “actual business value,” Kalishman declared.
He said AI vendors need to work harder when it comes to matching up their products to healthcare providers’ most pressing business and clinical needs.
“[AI companies] come in and say, ‘Look at this great feature — wouldn’t you love it?’ And we say, ‘Wonderful. What’s the clinical value? What’s the business value? How does it integrate into our EMR?’ And they don’t have great answers,” Kalishman explained.
AI companies need to have more of a plan when it comes to IT integration.
Kalishman said many of the AI vendors that he encounters don’t seem to understand “how truly messy” the data integration process is when a health system onboards a new AI product.
AI companies often don’t have an IT expert on their team who can answer health systems’ specific questions about what integration would look like, which is frustrating, he noted.
“There needs to be more of a technology infrastructure in place to adopt a lot of these AI technologies. Health systems’ data is messy. We have a lot of disparate IT systems all connected together, some of them more difficult than others to integrate into. We’ve got to figure that out to really take advantage of this AI,” Kalishman stated.
AI vendors should think about starting off with a discrete use case instead of offering a broad platform.
Another problem is that many AI vendors are pushing “these really broad products,” Kalishman remarked. He thinks young AI companies will have better luck in the beginning if they focus on selling “more discrete solutions.”
Regard, one of the AI companies in which Sentara has invested, is an example of a startup that is starting out with a discrete use case, Kalishman pointed out. The startup sells an AI co-pilot that helps clinicians diagnose medical conditions.
“If you believe the hype, generative AI has the opportunity to transform every piece of the health system business — from call centers to documentation to clinical decision support. A lot of companies are walking in and saying ‘We can solve multiple of these problems at the same time with our base platform.’ But with Regard, we started with inpatient hospitals with a focused product — one that solves a very specific need within our hospitals. And it works — it delivers an ROI. Starting with that discrete solution with hospitalists allowed us to really deliver an ROI for an AI product,” Kalishman explained.
Photo: metamorworks, Getty Images
