Compartmental multivariate analysis of exercise ECGs for accurate detection of myocardial ischaemia.

An accurate computer-assisted diagnostic method for detection of myocardial ischaemia, called MUSTA, is developed. MUSTA is based on compartmental multivariate analysis of variables available in the exercise ECGs, and is definitively implemented in Prolog. It is heuristically developed by determining diagnostic criteria, which interrelate a modified ST/HR-slope, ST-segment value and shape, and maximum heart rate, so that concordance with the TI-201 SPECT is maximised. In the learning group consisting of 47 patients, MUSTA provides a diagnostic accuracy of 98%, the detection of ischaemia being in absolute concordance with TI-201 SPECT. MUSTA is evaluated in a similar but independent group of 60 patients. Then, accuracy is 90%, and sensitivity is 94%. The performance characteristics are significantly better than those of the standard exercise ECG, whose diagnostic accuracy in these groups is 77% and 70%, respectively. This study suggests that MUSTA is a significant improvement for computerised assessment of myocardial ischaemia.