Can Machine-Learning Beat Alarm Fatigue?

Alarm fatigue can be a relentless contributor to physician burnout. Learn how burgeoning new tech can help us tackle the problem and hopefully end it once and for all.

Nurses and clinicians are waging a losing war against the thousands of constantly-ringing alarms that flood every moment with screeches and beeps.

Whether An IV fell out (it didn’t) or a patient’s heart stopped (it didn’t), these alarms are piercing, ever-present, and usually wrong. It’s no wonder nurses and doctors have lost their sense of alarm urgency.

So, how can nurses and doctors better separate the real alarms from the false positives?

What is Alarm Fatigue?

Nurses are required to be on guard at all times, ready to spring into action at the slightest alarm. However, most of the alarms that go off in a patient’s room are errors or false-positives — up to 90%, according to various studies. Even so, no alarm can be ignored — any one of them could be the alarm that saves a patient’s life and so all must be treated as if they’re life-threatening.

This sensory fatigue ultimately leads to more than just stress — it leads to more frequently missed alarms.

The Cause of Alarm Fatigue

According to the Joint Commission, alarm fatigue is caused by a few factors.

The most obvious being an alarm malfunction. Next comes incomplete training by staff — they may have put the electrodes or sensors on incorrectly, or in a manner as to easily fall off during normal circumstances.

Poor training can also lead to alarms being set too sensitively, or not being customized for the exact needs of the patient.

Gathering Sensor Data

The first step toward tackling alarm fatigue is to record all instances of alarms going off, including information like when the alarm went off, for how long, and why. Whether or not the alarm was justified can also be recorded.

Don’t worry about increasing paperwork, either — this data can be gathered without clinician input. For instance, when a patient lays on their IV line 17 times, a sufficiently advanced computer can tell the alarm is most likely a reoccurring error.

Machine Learning Reducing Alarm Fatigue

After data has been gathered and analyzed, machine learning can be used to make better decisions on when alarms go off.

Called “alarm suppression algorithms,” these machine-learning solutions have already been tested and proven to reduce false alarms. The idea is that advanced machine learning can gather and synthesize years of available data to determine normal working parameters for a sensor. It can then take incidents of false positives and compare them to false positives around the world.

Then, when a heart monitor or other sensor detects activity in the patient outside of its safe operating parameters, it can be run past the computer, which can decide if the event matches the symptoms of a false positive based on typical behavior.

The algorithm can then shut the alarm down and list it as an error.

What Can Staff Do?

Obviously, these algorithms are the most potent weapon in the fight against alarm fatigue.

However, that doesn’t mean that healthcare providers can’t take steps to improve the situation. First and foremost is to get decision makers to recognize how devastating alarm fatigue can be. Show them reputable stats that showcase how dangerous missed alarms can be. Form a committee or task force with members from every echelon of healthcare, from C-suite to in-the-field clinician.

The second step is to provide in-house training, not only on how to place and properly use all sensors and ECGs, but on how to manually change the parameters of sensors.

The method of alarm can also be customized. Does every alarm need to be audio? Can it be changed to a notification sent to a central nursing computer? Less dire alarms are perfect candidates for these changeovers.

The Infrastructure Needed to Relieve Alarm Fatigue

Of course, implementing machine learning and better training doesn’t happen overnight. A baseline of secure medical computers, improved network connectivity, and “Internet of Things” sensors are required to solve this pervasive form of work stress.

The better alarms are, the fresher doctors are, the better outcomes for patients everywhere. To learn more about the equipment you need to tackle alarm fatigue, contact Cybernet today.