Patient monitoring systems for critical care analyze a patient’s vital
signs
and should alert caregivers when the patient requires immediate attention.
The
lack of effective data integration and knowledge representation in these
systems limits the monitors’ utility to clinicians. Intelligent alarm
algorithms that use artificial intelligence techniques have the potential in
reducing false alarm rate and in improving data integration and knowledge
representation. Crucial to the development of such algorithms is a well-
annotated data set. In previous studies, clinical events were either
unavailable or annotated without accurate time synchronization with
physiological signals, generating uncertainties during both the development
and
evaluation of intelligent alarm algorithms. This project aims to help
eliminate these uncertainties by designing a system that simultaneously collects
physiological data and clinical annotations in real time, and to develop alarm
algorithms based on patient-specific data using this system.