AI can significantly reduce ICU false alarms, study finds

A three-part system using machine learning was able to greatly reduce the number of false alarms caused by bedside monitors in the intensive care unit, a new study shows.

Researchers from Massachusetts General Hospital in Boston developed a system to improve ICU alarms, as described in npj Digital Medicine. In some cases, nearly 90 percent of these alarms have been found to be false positives, resulting in "noise disturbance, desensitization and decreased quality of care," according to the study's authors.

The system was comprised of three stages: signal processing, to ensure proper annotation of vital signs; feature extraction, to distinguish between benign noise and alarm-worthy physiological signals; and optimized machine learning, to reduce the complexity of the bedside monitors' internal models and thus improve their accuracy in deciding whether an alarm is necessary.

The system was applied to a dataset previously used in the 2015 PhysioNet/Computing in Cardiology Challenge. The researchers achieved a score on the challenge higher than those of all other previously published methods. In conclusion, they wrote, "Such an approach therefore has the promise to improve the ICU environment for patients and healthcare providers alike."

More articles on AI:
AI helps physicians with no coding experience create automated tools: study
Google files patent for motion-tracking AI for 'intelligent' infant monitoring
Intermountain expanding use of AI for patient engagement in new Kidney Services department

© Copyright ASC COMMUNICATIONS 2019. Interested in LINKING to or REPRINTING this content? View our policies by clicking here.

 

Top 40 Articles from the Past 6 Months