Comparative accuracy of automated computer analysis versus physicans in training in the interpretation of electrocardiograms

We compared the interpretation of an electrocardiogram (ECG) made by computer with that made by physicians in training, as well as with the diagnosis made by cardiologists. ECGs were collected from 1058 Japanese adults (812 men and 246 women, mean age 49 +/- 19 years). With the computer program, the incidence of false-negative reports was 10.5% while that of false-positive reports was 16.5%, when compared with the physician's diagnosis. The incidence of a false-positive diagnosis with the computer was 18 times higher than that found by the physicians in training. The results show the advantages and the limitations in the use of computers for analysis of ECGs.