Automatic Sedation

Researchers develop algorithms to automate management of sedation in ICUs

(RxWiki News) Researchers have made advances towards being able to automate the sedation of patients in hospital intensive care units (ICUs), a step forward that may take a burden off of nurses while improving patient safety.

Currently, ICU nurses are in charge of a variety of tasks for numerous patients at the same time. Among these tasks is the responsibility of manual sedation, or the reduction of irritability or agitation using sedative drugs.

According to Wassim Haddad, a professor in the Georgia Tech School of Aerospace Engineering, manual sedation control is tedious, inaccurate, time-consuming, and occasionally of substandard quality.

In an effort to address this problem, a team of researchers have developed control algorithms that use clinical data to accurately calculate a patient's degree of sedation. The algorithms can also alert medical staff if any changes in the level of sedation are detected.

The hardest part about developing an automated system such as this, according to Georgia Tech postdoctoral fellow Behnood Gholami, is figuring out the variables that measure the degree of sedation of a patient. Knowing how to detect agitation (a measure of sedation) will allow an automated controller to provide a sufficient amount of sedation without going overboard or leaving the patient uncomfortable.

Through analyzing over 15,000 clinical measurements from 366 patients in intensive care units, the researchers classified patients as "agitated" or "not agitated." The measurements they used included patients' facial expressions, overall motor movement, response to a potentially noxious stimulus, stability of heart rate and blood pressure, non-cardiac sympathetic stability, and a nonverbal pain scale. The researchers also assessed patients' level of sedation using the motor activity and assessment scale (MAAS), a scale of zero through six.

The researchers found that their computer algorithms classified a patient as "agitated" in 18 percent of cases, while the motor activity and assessment scale (MAAS), score indicated that the patient was "not agitated." Conversely, the computer classified a patient as not agitated in five percent of the cases, while the MAAS score indicated that they were "agitated." Put more simply, these results show an 18 percent false-positive rate and a five percent false-negative rate.

Although this level of performance would take a significant load off of ICU nurses' responsibilities, it would in no way replace nurses as the primary mediator of the management of sedation, according to researcher James Bailey, the chief medical informatics officer at the Northeast Georgia Medical Center.

In order to improve the accuracy of their algorithms, the researchers will have to continue collecting clinical data from ICU patients in real time.

The researchers recently presented their findings at the IEEE Conference on Decision and Control.

Reviewed by: 
Review Date: 
February 15, 2011