A new study suggests artificial intelligence could spot signs of consciousness in comatose patients—well before doctors do. Researchers at Stony Brook University developed a tool called SeeMe, which uses AI to analyze subtle facial twitches in patients with brain injuries, per Scientific American. These micromovements, sometimes imperceptible to clinicians, can indicate a patient is trying to respond to commands like "open your eyes" or "stick out your tongue."
The study, published in Communications Medicine, involved 37 brain-injured patients who appeared unresponsive. SeeMe tracked facial movements down to the pore level as patients were given simple instructions. The AI detected eye-opening efforts in 30 of 36 patients and mouth movements in 16 of 17 cases where video quality allowed. Importantly, in several instances, SeeMe picked up on these responses days before doctors noticed any visible change—on average, 4.1 days earlier for eye movements and 8.3 days for mouth movements.
Lead researcher Sima Mofakham says the goal was to quantify consciousness using widely available technology, potentially offering a faster and more objective way to assess patients. Current neuroimaging tests are resource-intensive, time consuming, and not routinely used. Neurologist Jan Claassen, who wasn't involved with the study, notes that detecting consciousness sooner could influence treatment decisions and allow rehabilitation to start earlier, which has been linked to better outcomes. A separate study published in May found that "AI could enhance diagnosis, prognosis, and differentiation between states" of consciousness disorders.
,
uBlock 
,
and
Ghostery 
)
