A Cusum-based multilevel alerting method for physiological monitoring

Area of Science:

• Biomedical Engineering
• Physiological Monitoring
• Medical Device Technology

Background:

• Current physiological monitoring systems often generate false alerts due to artifact sensitivity.
• There is a need for more robust alerting mechanisms that can accurately detect significant physiological trend changes.

Purpose of the Study:

• To develop and evaluate a novel method for detecting changes in the direction of vital sign trends.
• To generate multilevel alerts based on the statistical significance of detected trend changes, reducing artifact-induced false positives.

Main Methods:

• Utilized one-point-ahead signal predictions based on exponentially weighted historical data.
• Employed two-sided cumulative sum (Cusum) of prediction errors tested against multiple thresholds for change point detection.
• Incorporated heuristic analysis of temporal signal shapes to determine alert triggers and certainty levels.

Main Results:

• The method demonstrated high accuracy in detecting trend direction changes for key vital signs: end-tidal carbon dioxide (90.2%), expiratory minute volume (89.4%), peak airway pressure (91.8%), and noninvasive blood pressure (95.4%).
• Algorithm-estimated certainty levels for true-positive alerts showed strong agreement with expert anesthesiologist evaluations.

Conclusions:

• The proposed method effectively detects changes in vital sign trends with high accuracy, outperforming traditional artifact-sensitive systems.
• This approach offers a more reliable multilevel alerting system for physiological monitoring, with validated agreement from clinical experts.