How CAIA is building a Cancer AI lab to enable cross-institutional research
CAIA’s eight pilot projects are powered by the hypothesis that breakthroughs in cancer research can come from studying the routine clinical data collected during every patient visit. A single lab result in isolation tells only part of the story. But by mapping routine data across more than a million cancer patients, researchers can track how clinical signals (like standard blood work) may shift over time. This can help researchers uncover hidden patterns to track disease trajectories, anticipate treatment responses, and forecast complications.
CAIA’s federated network is built upon de-identified electronic health records from leading cancer centers. Since AI models make better predictions when trained on larger and diverse dataset, CAIA’s Gen-1 dataset brings together structured clinical records from over one million patients across four NCI-designated cancer centers, providing researchers with secure, standardized data for training and AI modeling.
Accelerating cancer research using de-identified clinical data
In October 2025, when CAIA announced the launch of our federated learning platform, we also highlighted that our data infrastructure is powering CAIA’s initial set of eight pilot projects.
Researchers working with CAIA are focusing on the following areas:
Tracking disease trajectories: Creating specialized AI tools to monitor subtle shifts in health data and map disease progression over time.
Predicting treatment responses: Analyzing routine clinical data to better anticipate how patients will respond to specific therapies.
Forecasting complications: Identifying early signals in daily clinical encounters to forecast potential health events before they occur.
Over the past year, CAIA has transformed this robust data and technology infrastructure into an active, cross-institutional Cancer AI laboratory. To translate this foundation into active scientific progress, we needed a shared collaborative venue where researchers across institutions could connect and build together.
Bringing researchers together in a shared setting
Formal and informal lab meetings are common in academic settings. Similarly, bringing researchers from four independent cancer centers together creates a space to solve technical and scientific problems collectively.
The CAIA lab sessions offer researchers the opportunity to:
Solve shared technical barriers: Building models in a federated environment introduces specific technical challenges. The lab provides a forum for teams to share the challenges they encounter when deploying models and develop solutions that work across the entire network.
Refine federated workflows: Moving from local model development to federated infrastructure requires a shift in how scientific work gets done. In the lab, researchers and model developers have the opportunity to debug and optimize their work before scaling it across the broader federated network.
Get scientific feedback early and often: Lab sessions are a safe, informal environment to share early results and get critical but constructive feedback from people doing similar work. This approach also helps set the agenda and questions for subsequent lab sessions.
Fostering a scientific community and culture
Transitioning from single-site research to a true federated network requires more than just shared infrastructure. It requires a shift in how researchers work together across institutional boundaries.
The culture of the CAIA scientific community is built on the following values:
Open, informal collaboration: By holding confidential, conversational working sessions, researchers have a space to showcase early findings, discuss evolving challenges, and exchange insights that strengthen everyone’s approaches.
Collective problem-solving: Since the Alliance’s inception, teams have been challenged to approach CAIA’s activities with a “problem-solving mindset” — not just identifying potential barriers or issues, but actively proposing and searching for solutions. Teams approach complex technical and scientific questions with optimism and resourcefulness, focusing on how to advance the work together.
CAIA’s lab sessions include experts working with CAIA including clinicians, model developers, computational scientists, and data experts. At each session, scientific teams present their work in progress with a Q&A that enables collaborative discovery. At present, lab sessions take place at a monthly cadence.
By bringing researchers together to build federated AI models from routine clinical data within a Cancer AI lab environment, CAIA is establishing a scalable model for cross-institutional scientific discovery.
If you’d like to learn more about CAIA, subscribe to our newsletter for updates and follow us on LinkedIn and X.