A HYBRID COMPUTER SYSTEM FOR THE MEASUREMENT AND INTERPRETATION OF ELECTROCARDIOGRAMS

The electrocardiogram, probably more than any other record of a physiological variable, has been widely subjected to computer analysis.’-7 Several possible reasons for this are: the record represents an important parameter; only a simple transducer is required to convert it to a machine accessible form; the complexity of the record is great ehough to make computer analysis advantageous, yet simple enough to make computer analysis plausible, and good correlation exists between its wave-shape and the clinical condition of the heart. The purpose of the project described in this paper was to develop a method for measuring and interpreting the electrocardiogram (ECG) by computer through utilization of a pattern-recognition approach which would enable the program to interpret ECG’s of abnormal as well as normal rhythms. While it is possible to interpret nonarrhythmic ECG’s by examining a single heart cycle, arrhythmia interpretation requires the characterization of a consecutive string of heart cycles, as well as related conventional measurements. Such a characterization is produced by a method which requires analog pre-editing, selective averaging, measuring, monitoring, and interpreting. The nonarrhythmia measurement and interpretation portion of the program has been tested on over 500 patient records using the 12 standard leads. The results of these tests are included in this report.