Tag Archives: coma

Head, Technology

AI Can Detect Emergence From Coma Better Than Doctors

Leave a comment

Recovery from severe traumatic brain injury (TBI) can be frustrating for families and trauma professionals alike. It occurs in fits and starts, and the longest wait occurs while waiting for the patient to wake up. We perform serial neurologic exams and monitor closely for any visible response to commands.

A group of medical and engineering researchers at SUNY Stony Brook theorized that muscle movements in response to commands might be too subtle for human detection during early emergence from coma. They developed an AI system called SeeMe that studied the patients’ facial appearance down to the level of pores, and trained it to detect very fine motor movements in response to three commands:

  • Stick out your tongue
  • Open your eyes
  • Show me a smile

A group of 16 normal volunteers and 37 TBI patients were then tested while being video recorded. Both SeeMe and trained experts judged responses.

Here are the factoids:

  • SeeMe detected facial responses 4 days earlier than the trained experts
  • It detected eye opening in 86% vs 74% of human observers
  • SeeMe was able to detect mouth movements in 94% of patients without endotracheal tubes
  • The SeeMe-detected responses correlated with the clinical outcome at discharge

Bottom line: Severely brain-injured patients are able to respond to commands with subtle facial muscle movements before human observers can detect them. A specially trained AI like SeeMe can identify these movements and help predict recovery sooner than clinicians. Imagine being able to tell the family, who has been seeing their loved one making no progress, that improvement is occurring! And imagine what other applications focused AI can have on other clinical areas where human senses don’t have the capacity that carefully trained machines do!

Reference: Computer vision detects covert voluntary facial movements in unresponsive brain injury patients. Commun Med 5, 361 (2025). https://doi.org/10.1038/s43856-025-01042-y

CNS

Could There Be A Simpler GCS?

Leave a comment

The Glasgow Coma Scale (GCS) has been around forever. Or really, for about 45 years. It was actually developed in the early 1970s and known as the Coma Index. It was further refined into the GCS, when 1 was selected as the minimum component score. Ever since, it has been used as a common language among clinicians to communicate gross neurologic function and trends.

But it is still somewhat complicated. Oh no it’s not, you say? Then why do so many trauma resuscitation rooms have it posted on the wall? There are three components, each with a different number of possible values. And frankly, some are harder to remember than others. Decerebrate vs decorticate, right?

So what if someone told you that a single GCS component works just about as well as the whole bunch? Researchers have been piecing this together for years, focusing on the motor component of GCS (mGCS). There are two flavors of simplified score: mGCS and Simplified Motor Score (SMS). The mGCS is just what it sounds like: the full motor component of GCS, ranging from 1-6. The SMS is further simplified from the mGCS: mGCS of 1-4 tranlsates to SMS 0, mGCS 5 = SMS 1, and mGCS 6 = SMS 2. In my opinion, this is actually more complicated because you have to remember not only the 6 mGCS levels, but also the cutoffs to convert it to SMS.

Finally, a group from Oregon Health Sciences University in Portland performed a nice meta-analysis of the best individual studies.

Here are the factoids:

  • Only papers that compared total GCS (tGCS) to mGCS or SMS were included, and only if they analyzed a receiving operator characteristic curve. The statistics appeared sound.
  • tGCS was very slightly better than either mGCS or SMS at predicting:
    • in-hospital mortality
    • neurosurgical intervention
    • emergency intubation
    • severe TBI

Bottom line: Overall, the total GCS is slightly (just a few percent) better at doing the things listed above, compared to the motor score alone or the “simplified” (really?) motor score. Is this clinically significant in the field? Probably not. And its mere simplicity makes it appealing. 

However, there is one major problem to adopting the mGCS for use outside the hospital. Inertia. As I mentioned, we have been using the full GCS score for almost 50 years. Pretty much every trauma professional is familiar with the GCS or knows where to look up the details. But I suspect that those clincicians who assume care of the patient once in the hospital, and especially the intensive care unit (neurosurgeons) will never allow the use of an abbreviated scale. Good idea, but sorry, it won’t catch on in the real world.

General
Gallery Leave a comment

Nail Discoloration After Severe Traumatic Brain Injury (TBI)

Occasionally, patients who have had a severe brain injury but recovered relatively quickly may present with complaints of odd nail discoloration. This may involve fingernails and/or toenails. What gives?

This is actually a byproduct of repeated exams to determine the Glasgow Coma Scale score. A common way to determine the motor component is to squeeze the fingertip or toetip. I’ve seen some neurosurgeons use a pen to apply a great deal of force to the nail.

The discoloration is a resolving subungual hematoma. You may see different colors under different nails, depending on the age of the hematoma. Amaze your colleagues with your knowledge on this one!