Spaceflight isn’t easy on our bodies, and the eyes are no exception. A large proportion of astronauts experience vision problems, but researchers have now developed a new AI-powered detection tool to predict these issues before liftoff.
The US team focused specifically on spaceflight associated neuro-ocular syndrome (SANS), the term for vision degradation linked to time in space.
While symptoms of SANS can improve after astronauts return to Earth, that’s not always the case. Being able to identify who’s at risk could allow for preventative steps before missions even begin.
Related: Spaceflight Accelerates Aging of Human Stem Cells, Study Finds
“Our models showed promising accuracy, even when trained on limited data,” says ophthalmologist Alex Huang, from the University of California, San Diego (UC San Diego).
“We’re essentially using AI to give doctors a predictive tool for a condition that develops in space, before astronauts even leave Earth.”
That dataset was small because few people have been to space.
YouTube Thumbnail
The training was done using a supercomputer at UC San Diego, which developed a deep learning AI model that could spot patterns in eye characteristics that were associated with SANS. The data included samples from people who hadn’t been to space but had experienced some microgravity simulations.
To give the AI model enough data to train on, the researchers split eye scans into thousands of slices so they could all be analyzed and processed.
By the time the model was finished, it was able to predict SANS in astronauts with an accuracy of 82 percent, using pre-flight eye scans it hadn’t been trained on.
The data suggested SANS-related changes from spaceflight and from simulations were quite similar, so experiencing microgravity appears to produce similar changes to the eyes as real spaceflight.
“One of the most exciting findings was how similar the AI’s attention patterns were across both space and Earth data,” says ophthalmologist Mark Christopher, from UC San Diego.
“This strengthens the case for using Earth-based models to study space health – a promising development towards advancing human spaceflight research.”
The research also offers useful information about how and why SANS develops, through the areas of change the AI focused on, including the retinal nerve fiber layer and the retinal pigment epithelium at the back of the eye.
Though the researchers emphasize that their SANS detection system isn’t ready to be implemented yet, the overall aim is to improve astronaut health by spotting potential problems early, and developing ways to counteract them on missions.
If we’re going to spend extended periods of time in space and travel farther distances, we’ll need to address the associated health risks identified, which also include bone health, heart strength, and brain changes.
“The results and models from this study are early, but it’s a strong foundation,” says Huang. “With more data and refinement, this could become an essential part of astronaut health planning.”
The research has been published in the American Journal of Ophthalmology.