Why World Models Matter — Physical AI Comes to Medicine | Vivax
— by Vivax
The next leap in AI is physical: models that understand not just language, but the patient and the physical world around them.
For a decade, AI progress meant mastering language and images. That era is ending. The next leap is physical AI — systems that understand not just words, but space, motion, cause and effect, and the laws of the real world. NVIDIA frames this as the shift from generative AI to physical AI: models that can perceive, reason about, and act inside physical environments rather than only generating text on a screen.
At the heart of physical AI is the world model — a foundation model that learns how the world actually behaves, so it can simulate and predict what happens next. NVIDIA's Cosmos world foundation models do exactly this: trained on vast amounts of real and simulated data, they generate physically accurate scenarios and act as a kind of digital twin of reality, the same way large language models learned to predict the next word. Robotics, autonomous systems, and surgical platforms all need this shared understanding of the physical world to be safe and useful.
Vivax is building toward a unified clinical world model — one model that understands both the patient (their physiology, history, and clinical context) and the physical world around them (the operating room, instruments, imaging, and connected devices). This is the foundation of Acudx. The same world model powers our surgical assistant and our patient-facing clinical bot, and is designed to be embedded across every medical device we ship — so the AI reasons about the patient and the physical environment as one continuous, grounded picture rather than disconnected data points.