Waveform-to-patient matching across room transfers in the emergency department

Abstract

In the emergency department (ED), patients are often continuously monitored for physiologic deterioration. Signals used include electrocardiograms (ECGs), blood pressure waveforms, and respiratory waveforms. Due to the fast-paced and ambulatory nature of care in the ED, patient location is not always synchronized with waveform collection. The aim of our work is to link disparate waveforms belonging to the same patient. We utilize approximate admission and discharge times available from the ED administration database to delineate patient stays. We then use waveform-derived features of the ECG to link together sequential signals which are separated by variable time delays. The ultimate goal is to construct an openly available database of ED records with linked waveforms to stimulate research in emergency medicine.