Keynote Speeches

John Quackenbush portrait

John Quackenbush, Ph.D

Professor
Departments of Pediatrics and Human Genetics, University of Chicago
Emeritus; Professor and Chair in the Department of Biostatistics, Harvard University

Joel

Joel Saltz, M.D., Ph.D

Research Professor of Pathology and Laboratory Medicine
University of North Carolina at Chapel Hill
Emeritus; SUNY Distinguished Professor and founding Chair of Department of Biomedical Informatics, Stony Brook University

Joel

Yang Xie, M.D., Ph.D.

Inaugural David H. Perlmutter and Barbara A. Cohlan Distinguished Professor of Genetics
Founding Director, Center for AI-Driven Precision Medicine
Washington University School of Medicine in St. Louis

John Quackenbush full photo

Redefining Models of Health and Disease through Network Science

John Quackenbush, Professor, Departments of Pediatrics and Human Genetics, University of Chicago, Harvard TH Chan School of Public Health, and Channing Division of Network Medicine of Brigham and Women’s Hospital

Abstract: One of the central tenets of biology is that our genetics—our genotype—influences the physical characteristics we manifest—our phenotype. But with more than 25,000 human genes and more than 6,000,000 common genetic variants mapped in our genome, finding associations between our genotype and phenotype is an ongoing challenge. Indeed, genome-wide association studies have found thousands of small effect size genetic variants that are associated with phenotypic traits and disease. The simplest explanation is that genes and genetic variants work together in complex regulatory networks that help define phenotypes and mediate phenotypic transitions. We have found that the networks, and their structure, provide unique insight into how genetic elements interact with each other and the structure of the network has predictive power for identifying critical processes in health and disease and for identifying potential therapeutic targets. However, estimating the drivers of disease phenotypes or predicting therapies that might be useful is tremendously challenging—if not computationally intractable. Drawing inspiration from Wolpert and MacReady’s “No Free Lunch Theorem for Optimization,” we have found that biologically motivated constraints can help guide inference of biologically interpretable in scalable ways that provide meaningful insight into functional changes that can drive cancers and other complex diseases as well as how they are influenced by factors that include biological sex and age. Most importantly, the perspective enabled by this approach allows us to move beyond genetic variants alone to include regulatory controllers as major drivers of disease and health.

John Quackenbush, Ph.D, is Professor of Pediatrics and Human Genetics in the Division of Biological Sciences and the program in AI and Biological Complexity at the University of Chicago, a member of the University of Chicago Comprehensive Cancer Center, and Professor Emeritus at the Harvard T.H. Chan School of Public Health. John’s PhD was in Theoretical Physics, but a fellowship to work on the Human Genome Project led him through the Salk Institute, Stanford University, The Institute for Genomic Research (TIGR), Harvard in 2005, and UChicago in 2026. John’s research uses massive data to probe how many small genetic and other effects combine to influence our health and risk of disease. Key to his approach is modeling gene regulatory networks and understanding how these networks change between health and disease, over time, as a function of sex and gender, and between individuals. His more than 350 published papers have more than 106,000 citations, and his “NetZoo” software tools have tens of thousands of downloads. Among his honors is recognition in 2013 as a White House Open Science Champion of Change. In 2012, he founded Genospace, a precision medicine software company that was sold to Hospital Corporation of America in 2017. In 2022, he was elected to the National Academy of Medicine.

full photo

From Virtual Microscopy to AI-Mediated Discovery: Multiscale Cancer Biology and the Changing Practice of Science

Joel Saltz, Professor, Department of Biomedical Informatics, School of Medicine and College of Engineering and Applied Sciences, Stony Brook University

Abstract: Twenty-five years ago, virtual microscopy promised to transform routine pathology specimens into a resource for computational investigation of disease. For much of that time, pathology AI focused on reproducing the visual pattern recognition performed by expert pathologists. Today, advances in foundation models and scalable image analysis are enabling a much broader transformation. Routine H&E and immunohistochemistry images can be characterized across multiple scales, from subcellular nuclear features and individual cells to local cell populations, tissue architecture, tumor growth patterns, and whole-slide spatial organization. Increasingly, AI methods can also infer molecular phenotypes from routine histology. These characterizations create a reusable scientific resource for investigating biological and clinical relationships, generating hypotheses, and developing predictive models. From this broad exploratory foundation, selected measurements and models can be refined and validated to guide treatment selection, predict therapeutic response and clinical outcomes, and support the evaluation and rational design of new therapies.
Our research on tumor–immune interactions illustrates these opportunities. Integrating cellular composition, multiscale morphology, spatial organization, and molecular inference makes possible new quantitative descriptions of the tumor microenvironment. Collaborative studies, including work with investigators from the Carolina Breast Cancer Study, demonstrate how quantitative measurements of tissue organization and molecularly associated phenotypes can be related to clinical outcomes. Such studies are essential for evaluating the biological significance, reproducibility, and clinical relevance of new computational measurements.
Generative AI provides a complementary virtual laboratory. Conditional latent diffusion models can generate synthetic tissue images, including synthetic immunohistochemistry and rare morphological patterns. Beyond creating training data and educational materials, controlled generation provides opportunities to investigate whether learned models capture coherent relationships among cellular composition, tissue architecture, molecular characteristics, and spatial scales.
AI research harnesses are transforming how biomedical research is conducted by enabling investigators to delegate increasingly substantial and conceptually complex work to AI systems. What began with code generation is expanding to comprehensive computational investigations, experimental design, and collaborative research strategy. By connecting language models to persistent scientific context, data, computational tools, and rigorous evaluation, these harnesses allow scientific ideas to be implemented, tested, and refined iteratively. Drawing on our experience developing computational pathology methods with AI agents, I will discuss how this emerging partnership can help navigate the enormous space of hypotheses and analyses opened by multiscale biological and clinical data, while preserving reproducibility and keeping scientific direction and judgment in human hands.
In sum, these developments point toward two interconnected transformations: making biology computationally accessible and making scientific exploration computationally scalable. Their implications extend well beyond pathology, suggesting a new partnership between scientists and AI systems that may fundamentally reshape computational biomedicine.

Joel H. Saltz, MD, PhD, is a physician and computer scientist whose work connects advanced computing with biomedical research and clinical practice. His current focus is building Chilean Wool and Malachite, companies he co-founded in computational pathology and population health intelligence. He is Research Professor of Pathology and Laboratory Medicine at the University of North Carolina at Chapel Hill, helping lead cancer data science at the Lineberger Comprehensive Cancer Center. He also holds a research professor appointment in Biomedical Informatics and Computer Science at Stony Brook University.
His earlier computer science research helped make complex scientific applications run efficiently on parallel and distributed computers. His contributions included inspector-executor techniques for organizing computation and data movement, data-intensive systems such as Active Data Repository and DataCutter, and methods for managing spatial information. This work established foundations for analyzing large scientific datasets and later for integrating biomedical images with molecular and clinical data.
In digital pathology, his group and collaborators developed early virtual microscopy systems and methods connecting cellular and tissue features to molecular characteristics, treatment response, and patient outcomes. Their work advanced weakly supervised learning from whole-slide images and spatial analysis of tumors and tumor-infiltrating lymphocytes across cancer types. Open software has made these approaches more accessible and reusable.
In joint work with Stony Brook and Argonne National Laboratory collaborators, he has developed generative AI methods for synthesizing tissue images across magnifications, including coherent gigapixel whole slides. This work connects microscopic detail with tissue architecture and supports new approaches to creating training data and evaluating pathology AI.
Dr. Saltz founded and chaired biomedical informatics departments at The Ohio State University, Emory University, and Stony Brook University, building interdisciplinary programs that connect computer science, engineering, medicine, and biomedical research. He earned his MD and computer science PhD at Duke University and completed clinical pathology training at Johns Hopkins.
His clinical informatics research also uses electronic health records to study patients and populations. His current research explores how foundation models and AI agents can help researchers design, implement, and test new computational assays and analytical methods, integrating pathology, molecular, clinical, and population data to address questions in cancer and health.

John Quackenbush full photo

From Prediction to Partnership: Two Decades of AI in Precision Medicine

Yang Xie, Inaugural David H. Perlmutter and Barbara A. Cohlan Distinguished Professor of Genetics Founding Director, Center for AI-Driven Precision Medicine Washington University School of Medicine in St. Louis

Abstract: Precision medicine aims to deliver the right intervention to the right patient at the right time. Achieving this requires computational methods that turn complex biomedical data into actionable clinical insight. This talk will share experiences developing and validating data science and AI approaches across molecular, imaging, clinical and preclinical data, with applications in cancer prognosis, treatment selection and drug discovery. I will also discuss how AI is evolving from a predictive tool into a research partner, with large language models and AI agents emerging as collaborators in biomedical research. I will close by reflecting on the lessons that carry computational methods from development to real patient impact.

Yang Xie, M.D., Ph.D.,is the Inaugural David H. Perlmutter and Barbara A. Cohlan Distinguished Professor of Genetics; Founding Director, Center for AI-Driven Precision Medicine; BJC Investigator; and Siteman Cancer Center Pedal the Cause Professor at Washington University School of Medicine in St. Louis.
Dr. Xie develops computational and artificial intelligence methods to solve fundamental problems in precision medicine. Her research spans AI-driven prediction of therapeutic response and drug combinations, biomarkers for cancer treatment selection, integration of pathology imaging with genomics, and, more recently, large language models, multimodal foundation models and agentic AI systems for clinical and biomedical data. She has authored more than 250 publications, cited more than 25,000 times. Before joining WashU, she spent two decades at UT Southwestern Medical Center in Dallas. There she founded the Quantitative Biomedical Research Center, served as Associate Dean of Data Science and Founding Interim Chair of the Department of Health Data Science and Biostatistics, and held the Raymond D. and Patsy R. Nasher Distinguished Chair in Cancer Research. She is Deputy Editor of npj Digital Medicine and serves on the Advisory Board of Lancet Digital Health. Dr. Xie received her M.D. from Peking University and her Ph.D. in Biostatistics from the University of Minnesota.

Back to top