B. Heden et al., ARTIFICIAL NEURAL NETWORKS FOR THE ELECTROCARDIOGRAPHIC DIAGNOSIS OF HEALED MYOCARDIAL-INFARCTION, The American journal of cardiology, 74(1), 1994, pp. 5-8
Artificial neural networks are computer-based expert systems that lear
n by example, in contrast to the currently used rule-based electrocard
iographic interpretation programs. For the purpose of this study, 1,10
7 electrocardiograms (ECGs) from patients who had undergone cardiac ca
theterization were used to train and test neural networks for the diag
nosis of myocardial infarction. Different combinations of QRS and ST-T
measurements were used as input to the neural networks. In a learning
process, the networks automatically adjusted their characteristics to
correctly diagnose anterior or inferior wall myocardial infarction fr
om the ECG. Two thirds of the ECGs were used in this process. Thereaft
er, the performance of the networks was studied in a separate test set
, using the remaining third of the ECGs. The results from the networks
were also compared with that of conventional electrocardiographic cri
teria. Tbe sensitivity for the diagnosis of anterior myocardial infarc
tion was 81% for the best network and 68% for the conventional criteri
a (p <0.01); both having a specificity of 97.5%. The corresponding sen
sitivities of the network and the criteria for the diagnosis of inferi
or myocardial infarction were 78% and 65.5% (p <0.01), respectively, c
ompared at a Specificity of 95%. The results indicate that artificial
neural networks may be of interest in the attempt to improve computer-
based electrocardiographic interpretation programs.