Reducing Pulse Oximetry False Alarms Without Missing Life-Threatening Events

Alarm fatigue is one of the biggest problem in hospital environment nowadays, causing by excessive false physiologic monitor alarms. One of the possible reasons is due to the ineffective traditional threshold alarm system such as low blood oxygen saturation SpO2 alarm. We propose a robust two-stage classification procedure that can identify and silence false SpO2 alarms, while ensuring zero misclassified clinically significant alarms. Alarms and vital signs, which are related to SpO2 such as heart rate and pulse rate, within monitoring interval are extracted into different numerical features for the classifier. Multiple weak classifiers are implemented in the first stage and a specifically tuned support vector machines classifier is used in the second stage to make sure no life-threatening event will be missed. We evaluate the proposed classifier using a dataset collected from 100 participated children in the Children's Hospital of Philadelphia and show that the classifier can silence 26% of false SpO2 alarms without missing any clinically significant alarms.