Cluster and Discriminant Analysis in Myocardial Infarction Electrocardiogram

To study the clinical indexes of the myocardial infarction(MI) by the ECG, and find more effective techniques of the clinical diagnosis, 94 cases of the ECG records were selected, where 38 of which was acute myocardial infarction group(AMI), 11 of subacute myocardial infarction group (SMI), 19 obsolete myocardial infarction group(OMI), and 26 of which was control group. The following 7 indexes were collected from the 94 records respectively: (a)age, (b)gender, (c)chest pain, (d)diaphorase measurement, (e)abnormal Q wave, (f)elevational ST segment, (g)inverse T wave etc. Firstly, we found the classification of the above indexes according to the hierarchical clustering analysis, using the world famous statistics software SPSS. Then gave the discriminant function and the effect analysis according to the stepwise discriminant analysis. The results showed that: (1) all factors were classified into two categories: one was constituted of (a), (f),(g); another one was constituted of the rest 4 indexes; (2) the following 4 indexes were screened out as combined indexes of the discrimination function: (c), (d), (e) and (f), which coincided with the results Of the Clustering analysis. Finally the check Computation showed that, the overall discrimination sensitivity was 64.9%. and the discriminant sensitivities of the AMI, SMI, OMI. control were 57.9%, 54.5%, 47.4%, 92.3%. So we got the conclusion that the combined indexes constituted by (c), (d). (e) and (f), played an important role in the diagnosis of the MI.