Health system executives are in the midst of selecting the right AI vendor partners and projects for digital transformation. There are hundreds of pitches to sift through daily, and narrowing down the field of potential candidates is a time-consuming process.
Once the vendors are selected for pilots and then on to scaling systemwide, the IT executive’s job isn’t done. In fact, it’s just beginning.
“I’m really concerned about AI scope creep,” said Yasir Tarabichi, MD, Chief Health AI Officer at Cleveland-based MetroHealth and CMIO of Ovatient, during an interview with “Becker’s Healthcare Podcast.” “When we pick a solution for a particular pain point, or if it’s even a part of a larger platform, it starts to grow in terms of scope. First, it’s summarizing notes. Next, it’s listening to your conversation. But hold on, I already have an ambient AI solution. These things bump into each other, so I am worried about overall spend.”
MetroHealth is taking a measured approach to new technologies and pilot programs, favoring partners with flexibility. The constant new entrants and upheaval of market leaders means rushing to partner with the first company with a rigid agreement will leave health systems with an obsolete partnership before long.
“The biggest concern that I have continues to be spending money in the wrong places and on the wrong solutions,” said Dr. Tarabichi. “I still think we’re figuring things out, so we are treading lightly. We are trying to, at least when we commit to partners in the space, come in a way where we have leeway. Our EHR vendors are developing solutions and opportunities with generative AI specifically that start to conflict.”
Dr. Tarabichi knows the outcomes from a small pilot also may not scale to the larger organization. Many AI products promise to make clinicians’ lives easier, but it’s hard to quantify whether they are decreasing the providers’ cognitive burden or just changing the level of work they’re doing; instead of writing a note, they are reviewing an AI generated note and looking for hallucinations or misses.
“None of these solutions are perfect, so we have to be really careful about how much more work we’re throwing on our care team and making assumptions about whether or not we’re truly enhancing their workload,” he said.
MetroHealth is guarding against scope creep and faulty return on investment with a strong governance strategy bringing a diverse set of experts together. They’re also upskilling clinicians on what technologies can and can’t do, and empowering physicians to speak up when the AI isn’t working correctly.
The system is also keeping a close eye on the lifecycle of all solutions and decommissioning tools that aren’t working.
“There’s this natural assumption that more technology is always good, so we have to keep balancing that out to make sure we’re actually literally getting what we pay for, and making sure that we’re not making anything harder than it has to be,” said Dr. Tarabichi.
