NVIDIA GTC Healthcare: Physical AI, Surgical Robotics, and Isaac for Healthcare

This GTC session explains how physical AI is being applied to healthcare, focusing on NVIDIA's three foundational pillars—world models, robot policy models, and physics simulation—and their use in surgical robotics, hospital automation, and medical imaging. It also presents new datasets, foundation models, and deployment platforms for developers and researchers.

■ Foundations and Data
- Three pillars of physical AI: world foundation models, robot policy models, and physics simulation
- Open Age healthcare robotics dataset with 35+ partners, 11 embodiments, and 750+ hours

■ Surgical Robotics and Simulation
- Cosmos-based surgical transfer, predict, and simulator models, plus Isaac for Healthcare sensor simulation
- Post-training and evaluation of Groot H-based vision-language-action models

■ Deployment, Medical Imaging, and Examples
- Edge deployment with Holoscan 4.0 and IGX Thor
- Raw-to-insights for ultrasound and MRI, NV generate for CT/MR, and integrations with Hopper and Phillips
- Use cases from CMR Surgical, J&J MedTech, Moon Surgical, Peritas AI, and audience Q&A

This session is intended for healthcare robotics developers, medical device engineers, AI researchers, and technology strategists. Viewers can learn about NVIDIA's current healthcare physical AI platform, available open models and datasets, and the workflow from simulation to deployment.