Seattle Children’s Builds AI Assistant & Governance Model to Advance Pediatric Care

Dr. Clara Lin, VP and CMIO at Seattle Children’s, shares how the health system is deploying AI to support clinicians and transforming governance to drive IT decisions.

At Seattle Children’s, Dr. Clara Lin is leading an initiative that blends advanced AI with 15 years of carefully curated clinical pathways. The aim: to assist physicians at the point of care without compromising clinical rigor.

Developed in collaboration with Google, the AI-powered “Pathway Assistant” is trained exclusively on Seattle Children’s internal pathway documents, not a broad language model. It guides clinicians through diagnosis and treatment workflows for over 70 pediatric conditions, such as asthma or bruising, by engaging them in multi-step dialogues that adapt to specific patient details.

“If I asked it, ‘I have a four-year-old in front of me with asthma exacerbation, what should I do next?’ it doesn’t just give you an answer,” Lin said. “It will ask questions back to understand the patient’s condition before giving a recommendation.”

According to Lin, the assistant achieves more than 98% accuracy after repeated tuning and validation, and is being rolled out in emergency and urgent care settings. The tool does not yet connect to the organization’s EHR, partly by design. Operating independently of protected health information allows Seattle Children’s to avoid integration complexity while refining the tool’s performance.



“We chose to do it as a standalone,” she said. “If we decide in the future to share it with other organizations, we don’t have to worry about the integration part.”

This decision was also informed by the nature of the clinical pathways themselves. Many of the questions the tool asks are ones a physician would assess in real time, before documentation is even completed in the EHR. “A lot of that [clinical judgment] doesn’t exist in the EHR yet—it’s still in my brain as I’m trying to formulate a plan,” Lin explained.

While it isn’t intended as a diagnostic tool per se, the assistant is equipped to redirect a clinician if the information provided doesn’t match the intended pathway. “If it detects that your patient actually doesn’t belong to this pathway, then it will actually tell you that,” she said.

Some pathways contain embedded diagnostic steps. For example, the bruising pathway helps clinicians distinguish between anticoagulation issues, trauma, and potential abuse by guiding them through a series of context-driven questions.

Human-Centered Design and Informaticist Leadership

Lin, who still practices primary care pediatrics, says the success of the tool depends on a clinician-first approach to technology design. “That’s why informaticists exist: to be the liaison between clinicians and IT,” she noted.

Seattle Children’s CIO Zafar Chaudhry has cultivated a culture in which IT teams are encouraged to shadow clinicians and understand workflows firsthand. “He encouraged all of his IT workforce to be interacting with clinicians, to even round in the hospitals,” Lin said.

To further support clinical users, the organization created a tech support station known as the “tech pod,” located at the hospital entrance, where clinicians can receive help and reverse-demo their technical issues. Still, Lin emphasized that IT staff are not physicians or nurses, and cannot fully grasp the nuances of clinical care without continuous engagement from informaticists.

One of Lin’s signature achievements is the EHR Support Optimization Program (ESOP), which replaced a post-implementation group of physician superusers with a smaller, outcome-focused team. Under ESOP, 10 to 12 providers are funded annually for a modest portion of their time to complete one defined optimization project within their specialty.

“In the old model, we were spending a lot of money on expensive resources, but not really leveraging them,” she said. Now,