Reducing false asystole alarms in intensive care

Area of Science:

• Biomedical Engineering
• Critical Care Medicine
• Signal Processing

Background:

• High false alarm rates in intensive care units (ICUs) lead to alarm fatigue.
• Asystole alarms require immediate attention due to their life-threatening nature.
• Poor signal quality is a primary cause of false asystole alarms.

Purpose of the Study:

• To develop and evaluate an algorithm for identifying poor signal quality.
• To suppress false asystole alarms and improve patient safety.
• To reduce the burden of non-actionable alarms on critical care staff.

Main Methods:

• Developed an algorithm combining signal saturation and baseline wander with other physiological data.
• Trained and tested the algorithm on the MIMIC II and 2015 PhysioNet/Computing in Cardiology Challenge databases.
• Validated the algorithm on a separate neonatal dataset.

Main Results:

• Achieved 81.0% specificity and 95.4% sensitivity in detecting asystole alarms.
• Missed only one true asystole event out of 22.
• Successfully rejected 89.7% of false asystole alarms on a neonatal dataset while retaining all true events.

Conclusions:

• Basic signal quality evaluation can significantly reduce false asystole alarm rates.
• The developed algorithm shows promise in enhancing patient safety by reducing non-actionable alarms.
• This approach can help mitigate alarm fatigue in critical care settings.