Report: AI adoption by health systems is outpacing governance, infrastructure and strategy

Health systems across the industry are rapidly embracing AI solutions in administrative and clinical workflows, according to a new report from the Center for Connected Medicine at UPMC and KLAS Research. However, the study highlights critical gaps in strategy, validation, testing environments, and success measurement.

The report, based on a survey of more than two dozen health system leaders, indicates that while AI adoption has accelerated, many organizations are lagging in establishing the necessary infrastructure, governance frameworks, and strategic foundations for long-term success.

Rob Bart, M.D., chief medical information officer at UPMC, notes that the healthcare industry has swiftly moved from discussing AI's potential to actively deploying solutions across the enterprise. He emphasizes that implementation is merely the first step, and health systems must now build governance structures, testing capabilities, and organizational strategies to ensure AI delivers meaningful and measurable value.

The research reveals that despite near-universal predeployment testing, only 44% of health systems have dedicated testing environments, and 63% lack formal strategies. Resource, time, and talent limitations, along with inconsistent success metrics, pose challenges to governance, validation, and scalable implementation.

One example of addressing these challenges is UPMC's real-world data platform, Ahavi. Ahavi validates and improves third-party AI models using de-identified patient data to test AI solutions in silico before deployment, ensuring their impact on various applications without disrupting patient care.

Ken Howard, vice president at UPMC Enterprises, stresses that health care providers must invest in the right data infrastructure, governance processes, workforce capabilities, and evaluation frameworks to ensure responsible innovation when embracing AI technologies.