A knowledge base controlled signal processing and scoring system for the exercise ECG

A knowledge-based-controlled signal-processing and scoring system has been developed for processing exercise ECG studies. All of the processes from recording the analog ECG signal through calculation of the final scores are controlled by knowledge contained in two separate knowledge bases. Comparisons of differing processing and scoring methods are controlled by the user through editing of available knowledge bases. The system has been designed not only for evaluation of coronary artery disease (using accepted predefined scoring rules programmed into the computer) but also for research purposes, where user-defined rules may be entered and evaluated against current methods. The knowledge base contains knowledge regarding both measurements relating to specific segments of the ECG waveform and methods to score the information recorded during the exercise and recovery phases of a test. The authors also tested the system algorithms using different waveforms and various levels of noise. The average measurement error was smaller than the error measured by vision. For example, the average error of the J-point amplitude was about 0.1 mm