A computer-assisted monitoring system for arrhythmia detection in a medical intensive care unit

The arrhythmia detection capability of a computer-assisted monitoring system (CAMS) was studied in a large multidisciplinary ICU during an 18-month period. Four patient categories were evaluated: critically ill patients on mechanical volume respirators (group 1), patients with uncomplicated acute myocardial infarction (group 2), pacemaker-dependent patients (group 3), and patients on telemetry monitoring (group 4). ECG abnormalities were interpreted by the computer algorithm and recorded on paper. The same ECG abnormalities were analyzed independently by at least two critical care physicians unaware of the computer interpretations. The incidence of false-positive diagnoses (computer system errors) ranged from 10 in 1000 beats in groups 1, 2, and 4, to 20 in 1000 beats in group 3. Movement artifact accounted for 55.3% of all false-positive diagnoses. Of the total number of beats interpreted by the computer, 0.8% were false negatives and 3.8% were true positives. The most frequent true positive was pacemaker malfunction, which was diagnosed with 94% accuracy by the arrhythmia detection system. Significantly, rhythm abnormalities occurred as frequently in patients ventilated with mechanical respirators as in patients with acute myocardial infarction.