A new technique is developed thanks to artificial intelligence to better measure whether a patient is conscious or not

A new technique is developed thanks to artificial intelligence to better measure whether a patient is conscious or not

A small proportion of patients regain some consciousness during medical procedures, but a new study of brain activity could prevent such potential trauma. It could also help both comas and scientists trying to define which parts of the brain are key to the conscious mind.

Thanks to machine learning , it is now suggested that consciousness depends on the integration between the parietal cortex, the striatum, and the thalamus.

Most important brain regions

Integration measures, not just complexity, better detect changes in consciousness. The parietal / subcortical areas contribute more than the frontal areas to decoding consciousness. And the integration of the parietal and subcortical areas is a hallmark of conscious states.


All of this was what the UW-Madison researchers found by recording electrical activity in roughly 1,000 neurons that surround each of the 100 sites in the brain of a pair of monkeys at the Wisconsin National Primate Research Center over various states. of consciousness: under drug-induced anesthesia, light sleep, awake at rest, and awakening from anesthesia to a waking state through electrical stimulation from a point deep in the brain.

To select the characteristics that best indicate whether the monkeys were conscious or unconscious, the researchers used machine learning , an artificial intelligence technique, feeding their large data set to a computer.

They then told him what state of consciousness each pattern of brain activity had produced and asked which areas of the brain and patterns of electrical activity corresponded most strongly to consciousness. The results pointed in the opposite direction to the frontal cortex , the part of the brain that is generally monitored to safely maintain general anesthesia in human patients and the part most prone to exhibiting the slow waves of activity considered typical of unconsciousness.

As Michelle Redinbaugh , a graduate student in Saalman’s lab and co-lead author of the study , published in the journal Cell Systems , explains:

With data in multiple brain regions and different states of consciousness, we could put together all of these signs traditionally associated with consciousness, including the speed or slowness of the brain’s rhythms in different areas of the brain, with more computational metrics that describe how complex the brain is. what the signals are and how the signals interact in different areas.