Stanford Cancer Institute researchers are leading a multi-institutional initiative to integrate patient data across healthcare systems and cancer registries, allowing for enhanced cancer research. At the heart of this work is the Stanford-Sutter Health Oncoshare project, a growing database that links electronic health record (EHR) data from Stanford and Sutter Health with the California Cancer Registry (CCR), a comprehensive, state-wide repository for new cancers diagnosed in California that reports data to the National Cancer Institute (NCI).
The initiative is co-led by Stanford Cancer Institute members Summer Han, PhD, associate professor of neurosurgery and of medicine, and Allison Kurian, MD, MSc, professor of medicine and of epidemiology and population health. Together, they helm the Cancer Data Science Shared Resources Core within the Stanford Cancer Institute. Formerly known as the Biomedical Informatics Core, this shared resource now plays a central role in synthesizing clinical data for cancer investigators across the Stanford Cancer Institute.
“This core is providing many services for Stanford Cancer Institute members,” Kurian said. “We’re trying to provide harmonized and detailed data so they can conduct high-level, clinically impactful research.”
Painting a complete patient picture
This big data approach has enabled researchers to generate high-resolution maps of cancer treatment and identify care pathways that yield the best patient outcomes.
Unlike siloed EHR systems, which often lack continuity when patients transition between clinicians and healthcare systems, Oncoshare is designed to provide a comprehensive view of a patient’s cancer history.
“We don’t always know what happened before or after a patient’s time at Stanford,” Han said. “We need this full file of information for more rigorous research and long-term follow-up.”
Oncoshare addresses this gap by linking Stanford’s data to the CCR, and by expanding the reach through partnerships with neighboring health system Sutter Health.
Oncoshare Lung
This model, initially developed in 2008 for breast cancer under Kurian's leadership, has recently been expanded to lung cancer through a new NCI grant to Han.
“Allison had this model for a long time, but we didn’t have this model for lung. This was newly initiated one or two years ago,” said Han, who now serves as the principal investigator for the lung cancer Oncoshare project.
The lung cancer initiative has already begun generating significant early results. For example, lung cancer survivors receive CT scans for routine surveillance or if they develop a symptom, such as a cough, indicating the cancer may have returned. A recently published study found that algorithms designed to analyze images could accurately infer the reason behind lung cancer survivors’ CT scans, a detail that is often missed in real-world data.
“Real-world data is typically lacking information on why a CT was conducted after lung cancer diagnosis, leaving uncertainty about how CT surveillance impacts survival,” Han explained. “We used Oncoshare Lung, linked to the California Cancer Registry, and the model showed high discrimination with high probability. Based on this, we were able to see if people who received the surveillance scan had higher survivability than people who didn’t.”
Simply put, “If you know why the person received a CT scan, you can evaluate the potential benefits of surveillance,” said Han.
The Oncoshare Lung team holds monthly multidisciplinary meetings that bring together thoracic oncologists, surgeons, radiologists, data scientists, and other specialists to review and advance the work. A parallel monthly meeting is held by the Oncoshare Breast team. Together, they are building novel prediction models to assess patient prognosis, identify risk factors for disease recurrence and second primary cancers, and speed up chart reviews in the clinic when treating patients.
Future directions
But the vision doesn’t stop with breast and lung cancer.
“When Steve Artandi [director of the Stanford Cancer Institute] reached out to me and Allison to lead this new Cancer Data Science Core, he was hoping we could introduce the Oncoshare model to more of the Stanford Cancer Institute’s infrastructure,” Han said. “Can we expand it across all cancer disease sites, colorectal, brain, and so on, so more groups can benefit from this approach? The infrastructure is there, now it’s about growing the ecosystem.”
That expansion would also open access to Stanford Cancer Institute researchers more widely, providing members and trainees a direct line to specialized data.
With the Cancer Data Science Shared Resource Core providing support and infrastructure, Oncoshare is evolving into a foundational resource for Stanford Cancer Institute researchers, offering a powerful new tool for real-world, population-scale cancer research.
The power of population-based data is that it doesn’t leave anybody out. You get a sense of where everybody is in the spectrum of the population."
“The power of population-based data is that it doesn’t leave anybody out. You get a sense of where everybody is in the spectrum of the population. You don’t have myopia with a single institution,” Kurian said. “Being able to drill down into the depth of clinical detail that we have with electronic medical records, and even more now with AI tools, is the best of both worlds, and this is how we’re going to make big discoveries.”