AI detects 'covert consciousness' in unresponsive patients with severe brain injury

A new machine learning method analyzes routine electroencephalogram (EEG) recordings to identify previously undetectable cognitive function in patients who have become unresponsive following a severe brain injury.

The method is described in a study published June 27 in The New England Journal of Medicine. Researchers from New York City's Columbia University and New York University developed the machine learning model to detect brain activation in response to spoken commands given to clinically unresponsive patients in the first days and weeks after injury.

In the study, the model was tested on the EEG data of 104 unresponsive patients; as a result, brain activity was detected in 16 of the patients within a median of four days after their injuries. By the time of discharge, the conditions of 50 percent of those patients had improved to the point where they could follow commands, compared to only about a quarter of the unresponsive patients.

Furthermore, after a year of rehabilitation, nearly half of the patients who had shown signs of "covert consciousness" had achieved partial independence, compared to just 14 percent of the initially unresponsive patients.

Though it will be some time before the highly technical method becomes widespread in intensive care units, the study's findings mark a groundbreaking advancement in the care of patients with severe brain injury and are expected to eventually revolutionize the model of care for these patients.

More articles about AI:
Tech consortium including Google, Microsoft and more sets benchmarks for AI performance
1 in 5 consumers trust AI-generated healthcare information, survey finds
Brigham and Women's, Emory and more join radiology AI pilot program

© Copyright ASC COMMUNICATIONS 2021. Interested in LINKING to or REPRINTING this content? View our policies by clicking here.


Featured Whitepapers

Featured Webinars