Machine learning for early prediction of acute myocardial infarction or death in acute chest pain patients using electrocardiogram and blood tests at presentation

Table 3 Comparison of ESC 0 h and CNN-MB

	CNN-MB-RuleOut	CNN-MB-Intermediate	CNN-MB-RuleIn
Patients with 30d AMI or death (n = 185)
ESC 0 h-RuleOut	1	1	0
ESC 0 h-Intermediate	0	81	1
ESC 0 h-RuleIn	0	10	91
Patients without 30d AMI or death (n = 2194)
ESC 0 h-RuleOut	1035	86	0
ESC 0 h-Intermediate	273	738	5
ESC 0 h-RuleIn	0	29	28

Patients with or without 30d AMI or death are separated into two halves, and the number of patients ending up in the different groups according to ESC 0 h and CNN-MB are indicated by rows and columns. Numbers along the diagonal indicate that the models agree
AMI, Acute myocardial infarction; ESC 0 h, 0 h arm of the European Society of Cardiology algorithm; CNN-MB, convolutional neural network trained on median beat ECG data

ISSN: 1472-6947