Artificial intelligence in oncology depends on access to large, diverse datasets to build models that are accurate and generalizable, yet sharing sensitive patient data across institutions raises significant privacy, regulatory, and ethical concerns. Federated learning has emerged as a promising approach that allows AI models to be trained across multiple sites without the data ever leaving the institution, offering a potential path around one of the biggest barriers to collaborative cancer research.

But federated learning is not without trade-offs. While it offers stronger data privacy protections and enables institutions to contribute to model training without relinquishing control of their data, challenges remain around model performance, communication overhead, data heterogeneity across sites, and the complexity of implementation at scale. Federated learning may also not fully resolve privacy and security risks associated with models once they are trained; a distinction that increasingly matters as this technology moves toward real-world deployment. Understanding both the advantages and limitations of this approach is critical as the oncology community considers how best to harness AI while safeguarding the trust of patients with cancer and the institutions that serve them.

This webinar brings together experts with deep experience in AI model development and federated approaches in cancer research to explore the current landscape of federated learning in oncology. The speakers will examine current applications of federated learning in oncology, weigh the benefits against the practical and technical challenges, and discuss what the future holds for privacy-preserving AI in advancing cancer care.

The webinar will conclude with a live Q&A session led by Kenneth L. Kehl, MD, MPH.

Register here.