At Children's Hospital, AI Tackles Clinician Burnout by Making Data Actionable
Dr. Steven Martel, vice president and chief health information officer at Children's Hospital of Orange County, sees generative AI not as a silver bullet but as a tool that could finally make electronic health records work for clinicians instead of against them.
For decades, hospitals have built sprawling EHR systems meant to organize patient data. Instead, many clinicians say these systems bury them in documentation and administrative work. Chart reviews that once took an hour now take minutes at CHOC Children's thanks to AI tools that surface relevant information from structured and unstructured data in a single view.
The Real Problem AI Can Address
Clinician burnout stems largely from the cognitive burden of extracting actionable insights from massive databases built for billing and documentation, not clinical decision-making, Martel said. "When we speak about clinician burnout, in large part, this is due to the cognitive burden associated with the struggle to extract data that is clinically relevant and actionable for a specific patient."
The practical impact matters. For clinicians managing packed schedules and medically fragile patients, reducing a chart review from 60 minutes to several minutes changes the rhythm of a workday. Martel argues the benefit extends beyond speed: "We believe this creates a safer healthcare experience for our patients and improved EHR experience for our clinicians."
Skepticism Built In
Martel cautions against treating AI as a cure-all. Healthcare organizations need to understand how AI models are built and whether tools actually fit their patient populations.
"Understanding how these tools work, the underlying foundations of knowledge used to build their models, and careful assessment for applicability to your organization's patient population is critical," he said.
Data Culture Matters More Than Software
The bigger challenge isn't simply buying AI software. It's changing how hospitals think about data itself.
Successful AI deployment requires flexible data platforms and organizational tolerance for measured risk. Some initiatives will fail, and that's informative. Success often comes through repeated testing, refinement, and constraint-setting-not by flipping a switch.
"If leaders believe that the tools simply need to be 'turned on' and success will occur, they will be disappointed," Martel said.
Humans Remain Essential
As AI systems become more capable, Martel said healthcare leaders must keep patients safe and clinicians in control. Autonomous clinical decision-making should remain off limits.
"Autonomous workflows that involve clinical decision-making should be avoided," he said. "Preserving and operationalizing the human-in-the-loop framework for clinically related AI supported workflows is ideal to augment but not replace the clinical judgment of those delivering care."
The debate unfolding in hospitals now centers on trust-in the data, in the tools, and in the clinicians expected to use them. AI for healthcare works best when it supports human judgment rather than replacing it.
Your membership also unlocks:
