As clinical AI matures from research prototypes to real-world deployment, new challenges emerge around agentic systems, clinical validation, and human-AI collaboration. This session highlights cutting-edge work on autonomous AI workflows, prospective clinical evaluation, and the design of effective clinician-AI interfaces — addressing how AI can be reliably, safely, and equitably integrated into healthcare.
Machine and deep learning technologies are transforming our ability to analyze and identify patterns in complex, multidimensional medical datasets. The rapid evolution from static predictive models to agentic AI systems — capable of autonomously perceiving clinical data, reasoning across multiple sources, and executing multi-step workflows — marks a pivotal transition in how AI is being integrated into healthcare. Alongside these developments, large language models (LLMs), multimodal foundation models, and ambient clinical intelligence tools are moving beyond research prototypes and into active deployment within health systems. Leveraging these innovations has the potential to greatly enhance patient care by improving clinical decision-making, automating documentation, interpreting medical images, and optimizing triage and care coordination processes.
In this session, we invite submissions within the broad spectrum of emerging machine learning advancements that offer solutions to solve healthcare challenges. Our focus is on research areas that demonstrate how AI can address specific clinical needs. Submissions should involve either real-world deployment, clinical validation, or a clearly defined clinical application. Studies focused purely on molecular or omics-based methods without a direct clinical component are outside the scope of this session. We are particularly interested in papers that cover a variety of research topics, such as predictive analytics for patient outcomes, AI-driven personalized medicine approaches, natural language processing, federated learning, and LLMs for improved patient interaction and documentation, which showcase the power of collaborative AI model development while upholding data privacy and compliance and enhancing diagnostic accuracy. Our session will be dedicated exclusively to the clinical applications of these methodologies and excludes multi-omics methods that are well covered by other PSB sessions. Our goal is to promote discussions that explore how researchers in machine learning can collaborate with healthcare practitioners to enhance the efficiency and effectiveness of modern healthcare systems.
This session is interested in research on the applications of emerging artificial intelligence models in solving real-world and well-defined problems in healthcare, novel methodologies and unique applications of previously developed methods, and clinical implementation of artificial intelligence tools. Importantly, responsive submissions should include the deployment and/or evaluation of AI solutions in real-world healthcare delivery environments. Studies that focus solely on preclinical applications of AI or the secondary use of healthcare-related data without validating results in the clinic are considered off-topic and should be submitted elsewhere.
Below are examples of submission topics that would be of interest:
|
Dr. Fateme Nateghi Haredasht, PhD, is a postdoctoral scholar at the Division of Computational Medicine at Stanford University, where she is advancing machine learning integration in healthcare to unravel complex healthcare challenges and improve patient outcomes. |
|
Dr. David Wu, MD, PhD, is a physician-scientist in the Harvard Combined Dermatology Program at Massachusetts General Hospital. His research spans medical AI, precision medicine, and informatics. He led the creation of the First, Do NOHARM benchmark for clinical safety in language models. |
|
Dr. Kameron C. Black, DO, MPH, is an ABIM board-certified physician-scientist and Clinical Informatics Fellow at Stanford University. His research focuses on the safe deployment of agentic AI in real-world healthcare systems, including the first agentic benchmark for medical AI published in NEJM AI and featured by Stanford HAI, Forbes, and Anthropic. |
|
Dr. David JH Wu, MD, is a radiation oncology resident at Stanford whose research focuses on the intersection of artificial intelligence and clinical medicine. He is the creator and former host of the Medicine & Machine Learning (MaML) Podcast. |
|
Dr. Austin Schoeffler, MD, is a Clinical Informatics Fellow at Stanford Medicine, specializing in Emergency Medicine with a focus on clinical informatics and digital health innovations. His interests include clinical decision support, AI-driven workflow optimization, and digital health solutions for underserved populations. |
|
Dr. Dokyoon Kim, PhD, is an Associate Professor of Informatics in Biostatistics and Epidemiology at the University of Pennsylvania. As a Senior Fellow at the Institute of Biomedical Informatics and Associate Director of Informatics for Immune Health at the Perelman School of Medicine, Dr. Kim brings robust expertise in the integration of AI into translational informatics. He also serves as the Director of the Center for AI-Driven Translational Informatics (CATI). |
|
Dr. Brett K. Beaulieu-Jones, PhD, is an Assistant Professor of Medicine at the University of Chicago. His research interests center on the use of machine learning in biomedical data, with a focus on phenotype definition for complex conditions and the integration of artificial intelligence tools into healthcare settings. He is an organizer of SAIL and on the board of AHLI (ML4H and CHIL) and will advertise this session to these communities. |
|
Dr. Joseph D. Romano, PhD, FAMIA, is an Assistant Professor of Informatics and Pharmacology at the University of Pennsylvania. He is an expert in the integration and analysis of clinical and environmental health data using graph machine learning and other AI-based techniques and is a founding member of the NIH/NIEHS Environmental Health Language Collaborative. |
|
Dr. Geoffrey H. Tison, MD, MPH, is a practicing cardiologist, Associate Professor, and Co-Director of the Center for Biosignal Research at the University of California, San Francisco. He leads a computational research lab at UCSF (tison.ucsf.edu) that aims to improve cardiovascular disease prevention by applying artificial intelligence and statistical methods to large-scale medical data. |
Please see the PSB paper format template and instructions at http://psb.stanford.edu/psb-online/psb-submit.
Unlike the abstracts at most biology conferences, papers in the PSB proceedings are archival, rigorously peer-reviewed publications. PSB publications are Open Access and linked directly from MEDLINE/PubMed and Google Scholar for wide accessibility. They should be thought of as short journal articles that may be cited on CVs and grant reports.
Poster presenters will be provided with an easel and a poster board 32" x 40" (80x100cm); either portrait or landscape orientation is acceptable. One poster from each paid participant is permitted. See the submission portal website for the instructions regarding poster submissions.