The Clinical Automation & Data Driven Insights (CADDI) Lab at UCSF leverages clinical data, automation, and machine learning to improve patient care, clinical workflows, and operational efficiency in radiation oncology. We sit at the intersection of clinical practice and cutting-edge computational research.
We unite clinical practice, data science, and software engineering, translating research directly into deployed clinical tools.
We follow an iterative, clinician-in-the-loop development process, moving from raw clinical data to validated models to deployed tools that live inside real workflows.
The CADDI Lab organizes its research around four interconnected themes, each combining rigorous investigation, clinical deployment, and measurable impact. Together, they form a cohesive vision for the future of radiation oncology.

Machine learning models stratify plan complexity and predict patient-specific QA outcomes, enabling smarter and more efficient quality assurance in radiation therapy, and reducing reliance on time-intensive physical measurements.
Identify high-risk treatment plans before delivery
Intelligent QA platform serves as clinical decision support system
Prospective studies confirming clinical accuracy
Focus resources on the cases that truly need attention, not routine checks.
Catch high-risk plans proactively before they reach the treatment room.
Compare complexity to thousands of past treatment plans to understand drivers of risk of QA failure.
Our commitment to patient safety and treatment excellence drives our innovations in clinical automation. The CADDI Lab develops tools that streamline workflows, ensure consistency, and enhance the quality of radiation therapy planning and delivery.
Developed by Dr Evan Porter, automated treatment plan verification at multiple workflow hand-off points to identify potential risks and deviations from quality standards before treatment begins.
Custom-built scripts automate routine tasks, reducing manual effort, standardizing processes, and minimizing human error for improved efficiency.
Advanced tools visualize and compare dose distributions and organ-at-risk sparing across multiple treatment plans, empowering clinicians to make optimal treatment decisions.
Raw clinical data holds untapped potential. The CADDI Lab transforms data from Epic, oncology information systems, and departmental sources into real-time dashboards and actionable intelligence, empowering leadership and clinical teams to make decisions grounded in evidence.
Visual workflow analytics covering planning throughput, bottlenecks, and resource utilization.
Automated systems ensuring compliance metrics and department KPIs are captured and reported in real time.
A live whiteboard showing where each patient sits in the treatment workflow, surfacing which tasks are overdue, who is at risk of delay, and where bottlenecks are forming.
Robust integration with Epic and oncology information systems for structured data extraction.
Intuitive, interactive visualizations deployed across the department for clinical and operational use.
Automated notifications that flag when tasks become available or are completed, keeping the care team informed and accountable without manual follow-up.
The CADDI Lab builds scalable, reproducible data ecosystems that bridge the gap between clinical operations and translational research, enabling multi-institution collaboration and large-scale analytics that were previously out of reach.
A self-service research data platform empowering investigators to access and analyze structured clinical data independently.
Multi-institutional collaboration enabling pooled dataset analysis across leading radiation oncology centers, with harmonized data models for reproducible findings.
Adoption and development of oncology data models (O3) to standardize data representation across systems, sites, and studies.
Associate Professor and Director of Reporting & Analytics at UCSF Radiation Oncology. Dr. Witztum leads the lab's research agenda, clinical deployments, and academic partnerships, combining deep clinical expertise with a passion for computational innovation.
The lab works in close partnership with clinical faculty, medical physicists, and data scientists across UCSF Radiation Oncology, ensuring every tool is informed by frontline clinical experience and operational realities.
The CADDI Lab actively mentors UCSF and USF students, physics residents, and research trainees.
The CADDI Lab is seeking motivated, technically skilled trainees eager to work on problems that matter — real clinical systems, real data, and real impact. We offer a unique environment where research directly shapes patient care.
Join a team building and deploying automation tools in live radiation oncology workflows at UCSF.
A broader research internship spanning the lab's full portfolio — ideal for those wanting exposure to multiple domains of clinical AI and data science in oncology.
The CADDI Lab actively seeks partnerships across academia, industry, and clinical institutions. Whether you are a researcher exploring data collaboration, an organization with clinical deployment opportunities, or an institution looking to bring automation tools to your radiation oncology program - we want to hear from you.
Partnerships with UCSF Radiation Oncology, the University of San Francisco Health Informatics Program, and national and international research collaborators including BigROC consortium members.
We welcome engagement with companies seeking to validate, co-develop, or clinically deploy AI and automation tools in radiation oncology settings.
Institutions interested in adapting CADDI Lab tools for their own clinical environments are encouraged to reach out.

1825 4th Street, San Francisco, CA
CADDI Lab