Despite the rapid implementation of artificial intelligence solutions across the healthcare industry, many organizations are still working on establishing necessary infrastructures and governance foundations, a new report from UPMC’s Center for Connected Medicine and KLAS Research found.
Ninety-three percent of respondents report that their organizations deploy third-party AI solutions, though only 44% have a dedicated data platform or environment for tool testing. But 92% of respondents report testing third-party tools prior to deployment.
“What’s emerging from this research is a clear recognition that implementation is only the first step,” said Rob Bart, M.D., UPMC chief medical officer, in a statement. “Health systems are now focused on building the governance structures, testing capabilities and organizational strategies necessary to ensure AI delivers meaningful and measurable value.”
The “Validation and Trust: The Governance of AI Solutions at Health Systems” report drew insights from 27 healthcare leaders, including those from health systems and ambulatory care organizations.
Clinical documentation tools were the most commonly cited AI solution amongst respondents, with 52% reporting deploying such tools, followed by revenue cycle, coding and billing applications (36%).
Most respondents report data analysis is mostly within electronic health records (EHRs) or vendor analytics tools, followed by cloud data warehouses/lakehouses and multiple marts or warehouses. Five respondents reported being unsure.
The report notes that respondents say their “top pain points” for data quality include manual workarounds, spreadsheets and inconsistent definitions across different teams.
“This suggests that many health systems still rely on labor-intensive processes to reconcile, interpret, and prepare data for reporting or AI-related work,” the report said. “Missing or incomplete data, unstructured data, and timeliness issues are also commonly cited, underscoring the difficulty of turning large volumes of clinical and operational data into consistent, usable inputs.”
Analysts say the findings suggest AI-related challenges are “not only technical; they are also operational and governance-related.”
Top internal barriers for executing AI strategies systemwide include resource, budget, time or talent constraints; governance, security and compliance; change management, adoption and education; use-case/vendor selection and evaluation; return on investment (ROI) or financial cases and workflow redesign or operational engagement.
“Overall, the findings suggest that the next phase of AI strategy will depend less on enthusiasm and more on execution, including strong governance, clear problem statements, realistic resource planning, strong change management, and careful attention to how AI fits into daily work,” the report said.