Basic/Clinical Science Session
Teaching Machines Kidney Biology: Foundation Models That Learn Once and Generalize Everywhere
October 22, 2026 | 02:00 PM - 04:00 PM
Location: Room 603, Convention Center
Session Description
Artificial intelligence (AI) is increasingly used in medicine, but most tools are built for one single, narrow task at a time. Foundation models are different. They first learn the basic "language" of a system and can then apply it to many problems. In kidney research, a foundation model can learn core kidney biology by integrating genetics, single-cell and spatial data, pathology images, and clinical information. This session explains what foundation models and GPT-style AI are and why they may become a critical decision layer in kidney research, trials, and care.
Learning Objective(s)
- Describe what foundation models and GPT-style AI are, and explain how they differ from traditional single-purpose analytic tools used in medicine
- Explain how integrating genetics, single-cell and spatial biology, pathology images, and clinical data enables AI models to learn core kidney biology
- Identify practical applications of kidney foundation models for patient stratification, clinical trial design, and therapeutic target evaluation
- Describe current limitations, risks, and validation challenges associated with applying AI in kidney research and clinical care
Learning Pathway(s)
- Other
- Pathology
Moderators
Presentations
- What Are Foundation Models and GPT-Style AI, Really?
02:00 PM - 02:30 PM
- Pathology Foundation Models: Learning Kidney Structure and Injury
02:30 PM - 03:00 PM
- Single-Cell and Pharmacology Foundation Models: From Biology to Therapy
03:00 PM - 03:30 PM
- Radiology Foundation Models: Imaging as a Reusable Biological Signal
03:30 PM - 04:00 PM