A multivariate machine learning approach extracting 216 features from ABP and ECG signals achieved high accuracy for detecting false critical arrhythmia alarms in the ICU.
Does a machine learning methodology using ABP and ECG features improve the accuracy of false alarm detection for critical arrhythmia alarms in the ICU?
A novel machine learning approach using multivariate features from ABP and ECG signals can accurately detect false critical arrhythmia alarms in the ICU, potentially reducing alarm fatigue.
High false alarm rates in Intensive Care Unit (ICU) is a common problem that leads to alarm desensitization -- a phenomenon called alarm fatigue. Alarm fatigue can cause longer response time or missing of important alarms. In this work, we propose a methodology to identify false alarms generated by ICU bedside monitors. The novelty in our approach lies in the extraction of 216 relevant features to capture the characteristics of all alarms, from both arterial blood pressure (ABP) and electrocardiogram (ECG) signals. Our multivariate approach mitigates the imprecision caused by existing heartbeat/peak detection algorithms. Unlike existing methods on ICU false alarm detection, our approach does not require separate techniques for different types of alarms. The experimental results show that our approach can achieve high accuracy on false alarm detection, and can be generalized for different types of alarms.
Wang et al. (Tue,) conducted a other in Critical Arrhythmia Alarms. Multivariate machine learning approach extracting 216 features from ABP and ECG signals vs. Existing heartbeat/peak detection algorithms was evaluated on False alarm detection accuracy. A multivariate machine learning approach extracting 216 features from ABP and ECG signals achieved high accuracy for detecting false critical arrhythmia alarms in the ICU.