AI at VP&S Hosts Federated Learning for Health Workshop: Why, What, How?

Friday, May 16, 2025
10:00 AM – 3:30 PM

Roy and Diana Vagelos Education Center, VEC 401
104 Haven Ave, New York, NY 10032

This workshop aims to demystify federated learning (FL) and showcase its transformative potential in biomedicine and healthcare. From core principles to real-world applications, participants will explore how FL enables collaborative AI development across institutions, without compromising data privacy. We’ll examine case studies spanning medical domains, dive into ethical and regulatory considerations, and highlight emerging paradigms like personalized FL. The day will close with an industry panel and demo session. By fostering interdisciplinary dialogue, we aim to build a shared foundation—and spark momentum toward a federated learning infrastructure at Columbia.

This event is for clinicians, biomedical researchers, biomedical informaticists, computer scientists, data scientists, engineers, data privacy experts, IT and IRB leaders, and industry professionals. Whether you’re FL-curious or already federating in the wild, this is your chance to learn, challenge assumptions, and connect with others working to advance privacy-preserving AI in biomedicine and health.

Agenda

10:00 am - From Silos to Synergy: An Introduction to Federated Learning (Gamze Gursoy)
10:30 am - Case Studies on Federated Learning for AI in Ophthalmology: Thyroid Eye Disease, Macular Degeneration, and Glaucoma (Kaveri A. Thakoor and Angela McCarthy)
10:45 am - Personalized FL in Image-Guided Radiation Therapy (Yading Yuan)
11:00 am - Coffee Break
11:30 am - TBD (Despina Kontos)
11:50 am - TBD (Soojin Park)
12:10 pm - Cookie Monster: Privacy Architecture for W3C's Draft Standard on Privacy-Preserving Advertising APIs (Roxana Geambasu)
12:30 pm - Boxed Lunch
1:00 pm - Panel Discussion: Industry and Regulatory Perspectives (Representatives from Rhino Health (Ittai Dayan), NVIDIA (Holger Roth), J&J (Asha Mahesh), CTSA (Muredach Reilly), CUIMC IT (David Wentsler), CUIMC IRB)
2:30 pm - Demo Hour (Representatives from Industry and Columbia)

Organizers/Speakers

Gamze Gursoy

Title: Herbert Irving Assistant Professor of Biomedical Informatics, Columbia University; Core Member, New York Genome Center

Bio: Gamze Gürsoy is Herbert Irving Assistant Professor at the Department of Biomedical Informatics at Columbia University Irving Medical Center and a Core Member at the New York Genome Center. She is also affiliated with the Department of Computer Science. Her research group develops privacy-preserving tools to analyze and understand large-scale omics data in relation to diseases and phenotypes with a particular interest in developing software, file formats, and pipelines that enable broad sharing and analysis of sensitive genotypic and phenotypic data in public servers.

Kaveri A. Thakoor

Title: Assistant Professor of Ophthalmic Science (in Ophthalmology), Affiliate Faculty in Biomedical Engineering, Computer Science, Data Science Institute, Columbia University

Bio: Kaveri A. Thakoor, PhD, is an Assistant Professor in the Department of Ophthalmology at Columbia University Irving Medical Center. Dr. Kaveri Thakoor's Artificial Intelligence for Vision Science (AI4VS) laboratory is focused on transforming AI systems into teammates for ophthalmologists by tackling key challenges currently inhibiting the translation of AI to the clinic, such as robustness, interpretability, and portability. In her talk for this Federated Learning workshop, she will describe her team’s work toward developing more robust, generalizable AI-based ophthalmic disease detection systems by leveraging the power of federated learning to train on data from multiple sites while maintaining preservation of data privacy and model intellectual property.  She will discuss case studies applied to thyroid eye disease detection from facial eye images, macular degeneration detection from 2D OCT images, and glaucoma detection from 3D OCT volumes.  Dr. Thakoor and her research-year medical student, Angela McCarthy, will describe the regulatory and technical steps involved in implementing such federated learning consortia.

Yading Yuan

Title: Herbert and Florence Associate Professor of Radiation Oncology (Physics) (in the Data Science Institute) at CUMC

Bio: Yading Yuan, PhD, is the Herbert and Florence Associate Professor of Radiation Oncology (Physics) (in the Data Science Institute) at the Columbia University Medical Center, and the Director of Resident Program in Medical Physics (Therapy) in the Department of Radiation Oncology. His research lies at the intersection of computer engineering, physics and medical imaging, with a primary focus on clinical and scientific innovation in radiation oncology and on translating recent technical advancements in data science and engineering into clinical practice to improve patient care. His current research interests include large-scale medical image computing, multimodal predictive model, personalized federated learning and AI agents for radiotherapy.

Speakers

Despina Kontos

Title: Professor of Radiological Sciences (in Radiology) and of Biomedical Informatics, Columbia University

Bio: Despina Kontos, PhD, is a computer scientist with expertise in artificial intelligence, machine learning, and big data analytics for multi-modality imaging data. She is a professor of radiology and vice chair of artificial intelligence and data science research in the Department of Radiology at Columbia University Irving Medical Center (CUIMC) and director of biomarker imaging at NewYork-Presbyterian Hospital—with additional appointments in the Departments of Biomedical Informatics and Biomedical Engineering. She is also the founding director of Columbia University's Center for Innovation in Imaging Biomarkers and Integrated Diagnostics (CIMBID), a multidisciplinary center dedicated to developing and integrating quantitative imaging and non-imaging biomarkers for personalized disease prediction, particularly in cancer.

Soojin Park

Title: Associate Professor of Neurology (in Biomedical Informatics), Columbia University

Bio: Dr. Park is an Associate Professor of Neurology (in Biomedical informatics) at Columbia University Vagelos College of Physicians and Surgeons and Associate Attending Physician at NewYork-Presbyterian/Columbia University Irving Medical Center (NYP/CUIMC). Dr. Park received her undergraduate degree from Brown University and her medical degree from Drexel University College of Medicine. After completing her Neurology residency at Boston University Medical Center, she completed Stroke and Critical Care Neurology fellowships at the Massachusetts General & Brigham and Women's Hospitals.

Roxana Geambasu

Title: Associate Professor of Computer Science, Columbia University

Bio: Dr. Geambasu is an Associate Professor of Computer Science at Columbia University and a member of Columbia's Data Sciences Institute. Her research spans broad areas of computer systems, including distributed systems, security and privacy, operating systems, and applications of cryptography and machine learning to systems problems. Geambasu and her team seek a new model for how to address such personal privacy issues. She envisions a web environment where users are more aware of the privacy consequences of their online actions and make more informed decisions about the services they use. In her model, services and applications are held accountable for their actions and are explicitly constructed to protect user privacy. To forge this new web ecosystem, Roxana and her team design, build, and evaluate: (1) new transparency tools that increase society’s oversight regarding how applications use personal data in order to detect and deter unfair and deceptive practices; (2) new development tools that assist programmers in building applications that are privacy-preserving by design; and (3) new abstractions for responsible data management that promote and facilitate a more rigorous and selective approach to data collection and retention.

Panelists

Muredach Reilly

Title: Professor of Medicine, Director, Irving Institute for Clinical and Translational Research, Associate Dean for Clinical and Translational Research, Director, Cardiometabolic Precision Medicine Program, Columbia University

Bio: Muredach Reilly, MD, is the Herbert and Florence Irving Professor of Medicine. His lab is dedicated to translational and genomic studies of human cardiometabolic disorders. He is also director of the Irving Institute for Clinical and Translational Research. The Irving Institute’s mission is to improve human health by supporting efficient and effective clinical and translational research at CUIMC. The Irving Institute for Clinical and Translational Research, home to Columbia University’s Clinical and Translational Science Award (CTSA) Program hub and is one of over 50 medical research institutions across the nation that work together to speed the translation of research discovery into improved patient care. The CTSA hubs help to identify and overcome scientific and organizational barriers that slow the movement of discoveries from the lab bench to the patient bedside.

Ittai Dayan

Title: Co-Founder & CEO, Rhino Federated Computing, Columbia University

Bio: Dr. Ittai Dayan is the Co-Founder and CEO of Rhino Federated Computing, a company bringing federated AI to the enterprise level. His background is in academic medicine, machine learning and SaMD development. He previously held leadership roles at Mass General Brigham and Boston Consulting Group, and was a lecturer in Harvard Medical School. He published impactful papers on federated learning and has established partnerships with organizations including NVIDIA, cloud vendors, and major healthcare systems in order to progress the field. Under his leadership, Rhino has expanded the use of federated computing beyond healthcare to other regulated industries. Dr. Dayan remains engaged in industry collaborations and advisory roles, contributing to the development of AI applications that balance innovation with data privacy and security.

Holger Roth

Title: Principal Federated Learning Scientist, NVIDIA

Bio: Holger Roth specializes in developing distributed and collaborative software and models for various industries using federated learning and analytics. He has been exploring the topic both from theoretical and practical standpoints. During the COVID-19 pandemic, he led the experimentation of a federated learning study involving twenty hospitals around the globe to train more generalizable models for predicting clinical outcomes in symptomatic patients. His other research interests include computer-assisted annotation, active learning, and natural language processing. He is an Associate Editor for IEEE Transactions of Medical Imaging and holds a Ph.D. from University College London, UK. In 2018, he was awarded the MICCAI Young Scientist Publication Impact Award

Asha Mahesh

Title: Senior Director Janssen R&D Data Science & Digital Health, Privacy & AI Ethics at The Janssen Pharmaceutical Companies Of Johnson & Johnson

David Wentsler

Title: Information Systems Administrator, Department of Ophtalmology, Columbia University

Sponsors