AI Medic for Astronauts

What happens when an astronaut faces a medical emergency millions of miles from Earth, far beyond the reach of a real-time doctor? NASA is currently testing an offline artificial intelligence assistant called the Crew Medical Officer Digital Assistant, or CMODA, to solve this deep space peril.

The urgency of the project was highlighted earlier this year when NASA had to cut the Crew-11 mission short and bring them home from the International Space Station due to a medical concern. While an early return is possible from low Earth orbit, missions heading to the moon or Mars will not have that luxury, making local medical guidance a necessity.

To meet this need, Red Hat revealed that the cMoDA has transitioned from a cloud-dependent model into a fully disconnected edge deployment. The system is powered by an open-source tool called RamaLlama, which is backed by Red Hat to simplify how developers run and serve AI models.

According to Red Hat, Rama Lama provides the engine to run both large language models for complex medical reasoning and vision language models for image-based symptom analysis. This dual-engine setup allows the digital assistant to process both text and visual data without requiring a massive hardware footprint.

The entire system runs locally on the device, meaning astronauts can get critical diagnostics without needing any active connection to Earth. However, this space-age medical assistant has yet to actually leave Earth, creating a sharp contrast between its deep space design offline and its current terrestrial reality.

It is currently being run on a terrestrial twin of the HPE spaceborne computer, which is on its third iteration aboard the space station. Once the offline assistant is validated on Earth, developers will demonstrate the system to NASA leadership so they can evaluate its potential deployment into orbit.