Abstract
Recently, many studies have reported in which trials were made for applying digital computers to the pattern recognition of electrocardiograms (ECGs). It may be expected that an objective and quantitative interpretation of an ECG can be obtained quickly by the computer, provided that an appropriate program is set up. The process to reach a diagnosis would be followed exactly as written in the program by the computer which possesses remarkable properties such as a vast capacity of memory and a great accuracy and high speed in logical and analytical operations. However, one of the current problems in this field is to consider as to which one of different approaches is more adequate as the logical process in the interpretation of ECG's. In this paper, two diagnostic logics (or categorizers) currently in use were selected for comparison. One was derived from the principle of the logical decision tree and the other was from the principle of the joint probability. In this study, the discrimination of ECG's with or without myocardial infarction, as being considered to be important for the clinical use, was chosen as the subject to be tested for the advantages and disadvantages of these two categorizers as the diagnostic logic of the electrocardiographic interpretation. Material One hundred and one subjects of myocardial infarction (infarction) and 601 subjects without myocardial infarction (non-infarction) were selected. The non-infarction group consisted of 201 normals (Normal), 127 cases of left ventricular hypertrophy (LVH), 174 cases of right ventricular hypertrophy (RVH) and 99 cases of combined ventricular hypertrophy (CVH). The diagnosis of infarction was made on the bases of clinical findings, such as typical precordial pains lasting over one hour, the elevation of the activities of enzymes in the serum, such as glutamic oxaloacetic transaminase and lactic acid dehydrogenase, and serial ECG findings supportive of the history of an acute myocardial infarction. Individual conditions in the non-infarction group were diagnosed also on clinical findings, including history, blood pressure, 12-lead ECG's, X-ray films, phonocardiograms and the right heart catheterization.
