Impacting Cancer with HPC: Big Data, Bias and Digital Twins

Authors: Eric Stahlberg (Frederick National Laboratory for Cancer Research), Patricia Kovatch (Icahn School of Medicine at Mount Sinai), Jeff Buchsbaum (National Cancer Institute (NCI)), Thomas Steinke (Zuse Institute Berlin)

Abstract: High-performance computing has long been employed in cancer research and clinical applications. Yet recently, with the convergence of AI, large-scale computing and tremendous data assets, the role of HPC in cancer is undergoing a revolution. Borrowing from proven approaches in other HPC fields, the prospects for a personalized cancer patient digital twin, such as those comprising multi-scale integrated biological models are taking off. With implications for future clinical applications and research across the broader medical field, patient digital twins hold huge potential to be globally transformative to HPC and to medicine, as data and bias challenges are overcome.

Long Description: Cancer is a disease that touches us all. While HPC has long been used for research in cancer, the importance of HPC as a critical element of both cancer research and clinical applications has undergone a significant transformation in the past few years as large-scale computing, big data and machine learning have converged to bring new avenues for exploration. This annual BoF serves to bring the community together, both interested and curious, to explore the opportunities to impact cancer with HPC, and historically is well attended. The focus for this year’s BOF is the rapidly growing interest in biomedical digital twin capabilities, specifically focusing on the cancer patient digital twin as the exemplar that brings the community together. Drawing from proven uses of digital twin applications in engineering and other HPC sectors, the topic this year is expected to raise significant interest and draw many curious to learn avenues to bring their experiences to bear on this most challenging of diseases. As an open opportunity to the SC attendees for connecting with those employing HPC in the cancer community, this BoF serves as an essential connector and extension for the Computational Approaches for Cancer Workshop being held as part of SC conference. to build the community working in this area. The combined topics of data, bias and digital twins are critically important to the long-term success for HPC in cancer research and clinical applications. Overcoming challenges in data volumes, variety, sources, harmonization, disparities, and gaps are areas where HPC and related areas of AI and machine learning are expected to play a key role in future applications. Bias is essential to understand, characterize, and qualify to determine best use and make fit-for-purpose decisions in the growing ecosystem of data and predictive models employed to develop capabilities for cancer patient digital twins. Of particular interest is the challenge of bias in data based on population and demographic, where segments of the population are substantially underrepresented, which must be addressed for digital twins to reach general clinical adoption and adoption globally. The BoF is organized to provide an interactive forum to present these topics, framing challenges and opportunities in each to serve as a context for ensuing discussion and education. Experts with a variety of experience in these areas, including clinicians treating patients, will be part of the panel to provide a valuable perspective of how HPC is and can continue to impact individual patients. In its role to build the community of individuals bringing HPC to bear on challenges in cancer, the BoF will also include a brief synopsis of the most recent Computational Approaches for Cancer Workshop, while also highlighting opportunities and efforts to join with the community, which includes sharing new efforts started since SC20 for collaborative activities in cancer patient digital twins and precision radiation oncology involving both NCI and the US Department of Energy.

URL:


Back to Birds of a Feather Archive Listing