Seattle Children’s Hospital partnered with Google Cloud to launch Pathway Assistant, a generative AI tool that helps clinicians access over 70 clinical care pathways in seconds. Built on Google’s Gemini model via Vertex AI, the assistant streamlines decision-making, reduces administrative burden, and improves consistency in pediatric care.
What Is Pathway Assistant?
Pathway Assistant is an AI-powered tool developed by Seattle Children’s Hospital in partnership with Google Cloud. It uses generative AI (Gemini, via Vertex AI) to help clinicians instantly retrieve and interpret standardized care pathways for over 70 pediatric conditions.
Rather than searching through PDFs or relying on memory, clinicians can ask natural-language questions like, “What’s the care protocol for a 3-year-old with asthma exacerbation?” and receive a citation-backed, conversational response in seconds.
Why Was It Created?
Seattle Children’s has spent more than a decade developing Clinical Standard Work (CSW) protocols—evidence-based, condition-specific pathways to ensure high-quality, standardized care. But despite their value, these protocols were buried in static documents.
Until now, accessing them took up to 15 minutes—too long in a busy clinical setting. Pathway Assistant changes that, giving providers quick access to the right guidance at the right time.
How Does the AI Assistant Work?
- Data Source: It pulls from the hospital’s vetted Clinical Standard Work protocols.
- AI Engine: Uses Google’s Gemini large language model, fine-tuned for pediatric care.
- Functionality: Interprets structured and unstructured content—like flowcharts, images, and clinical notes.
- Dialogue-Based: Responds to clinician queries with follow-up questions, clarifying needs and providing tailored responses.
- Citation Support: Points to specific sections in pathways, reducing liability and increasing trust.
What Are the Benefits?
| Feature | Benefit |
|---|---|
| Fast Retrieval | Reduces search time from 15 minutes to seconds |
| Contextual Accuracy | 98%+ accuracy in clinical pilots |
| Evidence-Based Guidance | Pulls directly from Seattle Children’s CSW database |
| Clinician-Led Design | Developed and tested by frontline medical staff |
| Reduced Cognitive Load | Minimizes mental overhead during decision-making |
| Conversational Interface | Supports natural-language queries and interactive dialogue |
Is It Safe and Secure?
Yes. The assistant was built with privacy and safety in mind:
- All interactions occur within Seattle Children’s secure environment.
- No patient data is shared with Google.
- The tool underwent a “fail-fast” testing process to avoid hallucinations before launch.
- Only vetted pathway content is used—no external web sources are involved.
How Is It Being Used Now?
During the pilot phase:
- 40–50 clinicians tested the assistant across a range of conditions.
- Feedback shaped the model’s tone, confidence level, and user interface.
- It is now live across departments and expected to expand to include more specialties and use cases, such as telehealth triage, oncology, and emergency medicine.
Why Did Seattle Children’s Choose Google?
According to Dr. Zafar Chaudry, Chief Digital and AI Officer, Google was selected for its:
- Enterprise AI tools (Vertex AI)
- Support for custom fine-tuning
- Collaborative approach with clinical teams
- Focus on ethical AI development and healthcare compliance
What’s Next?
Seattle Children’s plans to:
- Expand the assistant’s knowledge base
- Integrate it more deeply into clinical workflows
- Evaluate its long-term impact on provider burnout, care consistency, and patient outcomes
What This Means for the Future of AI in Pediatrics
Pathway Assistant offers a working model of AI-enhanced clinical decision support that doesn’t replace physicians—it empowers them. With speed, clarity, and trust built in, this tool exemplifies the best of what generative AI can do for healthcare: help human experts do what they do best—better.
For MedPulse AI readers, the message is clear: we are entering a new phase of clinical care where AI agents act as invisible teammates, reducing friction, surfacing evidence, and ensuring that care is delivered not just faster, but smarter.
