By tracking brain activity through electroencephalography, or EEG, software may be able to maintain a patient in a medically induced coma more safely than a human expert can.

Anesthesiologists use EEG to monitor a patient’s level of sedation through sensors placed on the scalp. When a patient is deeply sedated in a medical coma—a technique sometimes used to reduce brain swelling after a traumatic injury or to treat uncontrolled seizures—a nurse or doctor must currently monitor the patient’s brain activity and adjust the rate of anesthetic delivery around the clock, sometimes for days.

Emery Brown, an MIT neuroscientist and an anesthesiologist at Massachusetts General Hospital, thinks the computer-controlled anesthetic system he has developed could do a better job. In a study published on Thursday in PLoS Computational Biology, Brown and colleagues demonstrate the technology in rats as a step toward developing it for human patients.

The potential for computer-assisted sedation stems from extensive work researchers have done to understand and control brain states during anesthesia. In recent years, brain monitoring technologies such as EEG and MRI have helped begin to unravel the differences between conscious and unconscious brains, says Martin Monti, a cognitive psychologist at the University of California, Los Angeles, who was not involved in the new study. Such work could help answer basic questions such as whether multiple brain functions are necessary to produce consciousness and whether loss of consciousness after severe injury is similar to unconsciousness during sedation or sleep, he says.

Brown’s group has been studying the anesthetized brain both to further scientific understanding of consciousness and to make anesthesia safer and more effective (see “The Mystery Behind Anesthesia”). The pattern of brain activity that doctors monitor to control sedation is well defined and can be recognized by a computer, says Brown. In fact, he says, the computer can be more accurate than the human eye at spotting how a patient’s activity pattern differs from the one that’s ideal for sedation, and it can make adjustments without under- or over-shooting the amount of drug required to maintain the sedated state. That could help ensure that patients aren’t given more anesthetic than they need.

The system could potentially be adapted to target the well-defined EEG signatures associated with levels of sedation used during surgery or in other situations, he says.

Other groups are exploring the possibility of computer-assisted sedation as well. Johnson & Johnson has developed a system to automate partial sedation of patients being screened for colon cancer. Sedasys was approved by the FDA in May, and J&J says it will begin selling the system in early 2014.

Automated sedation could be helpful to anesthesiologists or intensive-care nurses, says Mark Newman, an anesthesiologist at Duke University. However, he notes that monitoring a patient in a medically induced coma requires much more than EEG—heart activity and kidney activity, for instance, must also be tracked. So the technology could improve the precision of sedation, but it couldn’t automate it entirely.