Why AI matters at this scale

The Division of Biomedical Informatics (DBMI) at UC San Diego is a premier academic research unit focused on creating knowledge from complex biomedical data to improve human health. It operates at the intersection of computer science, medicine, and biology, developing methods and tools for data-driven discovery. As part of a massive R1 university and integrated health system, its work spans genomics, clinical informatics, population health, and imaging.

For a large academic division within a major research university, AI is not a novelty but a core strategic accelerator. At this scale—with hundreds of researchers, clinicians, and staff—the division manages petabytes of sensitive data. AI is essential to extract signal from this noise, enabling breakthroughs at a pace impossible through manual analysis. The scale justifies investment in dedicated AI infrastructure and talent, positioning the division to lead in competitive federal grant landscapes and industry collaborations. Failure to adopt AI risks obsolescence in a field where data complexity is growing exponentially.

Concrete AI Opportunities with ROI Framing

1. Automated Clinical Phenotyping from EHRs: Manually defining patient cohorts for research is slow and error-prone. An NLP pipeline to extract phenotypes from unstructured clinical notes can cut cohort identification time by over 70%, directly accelerating dozens of research projects and increasing grant output. The ROI is measured in additional grants secured and faster time to publication.

2. AI-Augmented Genomic Diagnostics: Interpreting genomic variants is a bottleneck in precision medicine. A deep learning model that prioritizes pathogenic variants could reduce analysis time per case from hours to minutes. This increases the throughput of diagnostic labs, potentially generating new clinical service revenue and improving patient outcomes, which enhances the medical center's reputation and attracts referrals.

3. Predictive Analytics for Hospital Operations: Leveraging admission and operational data to forecast patient flow and readmission risk. A successful model could optimize bed allocation and reduce preventable readmissions, directly saving the health system millions annually. The division can showcase this as a tangible return on its informatics research, strengthening internal advocacy and budget justification.

Deployment Risks Specific to This Size Band

Large academic divisions face unique AI deployment risks. Data Governance and Compliance is paramount; a single privacy breach involving PHI can result in massive fines, loss of funding, and reputational damage. Integration Complexity is high due to sprawling, often siloed legacy systems across the university and health system. Deploying a model at scale requires navigating diverse IT policies and stakeholder buy-in. Finally, Talent Retention is a constant challenge. With intense competition from industry for top AI/ML scientists, the division must offer compelling research problems and competitive resources to retain the expertise needed to execute its AI vision. Managing these risks requires dedicated project management, robust legal/contracting support, and clear communication of AI's value to institutional leadership.