A hallmark of NCI-designated cancer centers is that the science conducted at these institutions meets rigorous standards for transdisciplinary science and state-of-the-art research. The resources and computing environments at these institutions encourage multidisciplinary teams to creatively solve hard problems and ensure solutions are implemented at a relatively fast pace. The NCI would like to leverage these scientifically rich environments and invites teams of biomedical informaticians, data scientists, clinical researchers, and others to establish proof-of-concept federated learning frameworks to run multimodal AI models in coordination with other Cancer Centers and investigators at the NCI.
Federated Learning (FL) models enable researchers to test and train artificial intelligence (AI) models while preserving the privacy of the patient’s data. FL is a collaborative AI approach in which training data is not centralized and stays within organizational boundaries. These boundaries preserve the privacy of the individual contributors. FL models enable the research algorithms to be shared among institutions, not the data, thereby ensuring patient privacy and trust
Recently, researchers have begun focusing on multimodal AI approaches to incorporate imaging, genomic, and clinical data into their models to address the issue of modeling complex diseases. Instead of relying on data from a single modality, such as clinical factors, genomics, or radiology imaging, multimodal AI combines attributes from many approaches and uses them to describe the disease. Please see Acosta, et. al. for a current review of the field. This means the description used to predict patient outcomes is more data-rich, complete, and nuanced.