Cedars-Sinai Bets on ‘Data-Driven Intelligence’ to Fuel AI Revolution in Healthcare
Table of Contents
Cedars-Sinai is strategically positioning itself to capitalize on the rapid advancements in artificial intelligence, but with a crucial focus on maintaining clinical workflow, robust governance, and scalable infrastructure, according to a recent discussion with Mouneer Odeh, VP and Chief Data and AI Officer. The health system’s approach frames data and AI not as separate entities, but as a unified continuum designed to accelerate research, enhance patient care, and improve operational efficiency through what Odeh terms “data-driven intelligence.”
The Critical Role of Data Quality
Odeh emphasized that the success of any AI implementation hinges on the quality of the underlying data. “The difference between AI that behaves like a realy good graduate student or a fantastic assistant, and AI that behaves like your drunk friend is quality of the data,” he stated. This conviction underscores the organization’s commitment to data governance as a foundational element of its AI strategy, ensuring reliability and minimizing potential hazards. Strong linkages with clinical informatics, research leadership, applications, infrastructure, and security are also paramount to ensure models are deployed effectively within real-world clinical settings.
A Tripartite AI Portfolio: platform, Best-of-Breed, and In-House
Cedars-Sinai’s AI investments are currently distributed across three key areas: internal development, platform-delivered capabilities – particularly within the Electronic Health Record (EHR) – and targeted acquisitions of best-of-breed solutions. Odeh explained that this diversified approach allows the organization to leverage existing infrastructure while together pursuing innovative solutions tailored to specific needs. He highlighted the importance of usage measurement, and iterative improvement based on real-world performance. He argued that strong customer-relationship skills within IT are now as vital as technical expertise, as trust and responsiveness are essential for expanding successful pilot programs.
Governance is central to this evolution. Multidisciplinary bodies are being established to determine data access, define expected actions, interpret results, and establish criteria for model “publish-readiness.” Informatics experts play a critical role in bridging the gap between clinical realities and technical integration, ensuring models are deployed effectively to the right users at the right time. Transparent communication about a tool’s capabilities and limitations is also key to building trust and fostering iterative improvement.
Democratizing AI Through ‘Prompt-athons’ and Learning Communities
Cedars-Sinai is actively working to broaden AI expertise across the organization through “prompt-athons.” These training sessions empower non-IT teams – including HR, supply chain, patient experience, and performance improvement – to build task-specific agents using retrieval-augmented generation (RAG) grounded in Cedars-Sinai policies and playbooks. The most promising solutions are then refined through quality assurance and made available for wider use. This model is expanding to include IT and revenue cycle teams, fostering learning communities that accelerate technique diffusion and reuse.
These communities are supported by functional champions who ensure the long-term viability of solutions, keeping use cases current and content authoritative. Odeh emphasized the importance of celebrating successes to encourage participation and the need for a measured approach – consistent nurturing and governance are essential for compounding the benefits of these initiatives.
Key Takeaways for Scaling AI Initiatives
Odeh outlined several key principles for organizations navigating the complexities of AI implementation:
- Treat data governance as model governance; quality and lineage are safety rails for agentic workflows.
- Balance platform convenience with targeted bets on best-of-breed and internal builds where they create distinctive value.
- Move from project to product: measure usage, iterate in production, and sustain customer relationships.
- Make workflow integration the default; deliver outputs where clinicians already work and think.
- Stand up multidisciplinary governance that decides audience, actions, and publication readiness.
- Build learning communities: train non-IT teams,publish hardened exemplars,and cultivate functional champions.
- Communicate transparently about limitations and roadmaps to maintain momentum and trust.
Ultimately, Odeh’s guiding principle for peers is simple: “Think big, start small and scale fast.”
