AI Could Be the Answer to Alarm Fatigue
Patient monitoring technology has proven both a boon and, in some ways, a burden to medical science.
Automatic sensors can detect heart fluctuations, sounding an alarm that will bring staff running. Other sensors monitor respiration, brain activity, temperature and multiple other critical factors, alerting professionals when help is needed.
However, these monitors can be so sensitive they send alerts for even minor departures from preset norms. A busy surgical unit can be a noisy place with alarms going off so frequently for little reason that they become part of the background ambiance, causing what medical professionals call “alarm fatigue.”
In one large study, nurses in a busy urban hospital were bombarded by an average of 187 alarms per bed each day. Of the 2,558,760 alarms recorded during the month-long study, most – up to 95% -- were false or of little consequence.
So serious is alarm fatigue that in 2013 The Joint Commission issued a Sentinel Event Alert warned about the potential for desensitization. “In response to this constant barrage of noise, clinicians may turn down the volume of the alarm, turn it off, or adjust the alarm settings outside the limits that are safe and appropriate for the patient – all of which can have serious, often fatal, consequences.”
Still listed as one of the “Top 10 Health Technology Hazards” by the Emergency Care Research Institute, there is hope that yet another technological advance may hold the solution to too many alarms.
At Johns Hopkins, the health system’s alarms committee has been using and testing a number of techniques for quieting unnecessary alarms. Among these is the use of algorithms to decide when to sound an alarm, to whom and when and how to escalate the situation.
A more extensive use of artificial intelligence was discussed last fall in the Journal of Medical Internet Research. Researchers tested their AI algorithms against the recorded data from 32 surgical patients in Australia. Their technology reduced the total number of alarms by 99.3%.
Although it was not used in an actual clinical environment, “The experimental results strongly suggest that this reasoning algorithm is a useful strategy for avoiding alarm fatigue,” they wrote.
Using artificial intelligence to decide when and how to sound an alarm is still in the future. But, notes The Medical Futurist, “With time, AI solutions will be incorporated in patient monitors as a built-in “smart alarm system” throughout hospital units.”
Image by Bokskapet