Comparison of different statistical approaches for urinary peptide biomarker detection in the context of coronary artery disease

Table 6 Diagnostic/prognostic performance of classifiers modelled by SVM

Biomarker patterns	Biomarker number	CADD	AMIP		ACD
Biomarker patterns	Biomarker number	CADD	0-5 y	5-11 y	0-5 y
CADD t-score + AMIP all BM	123	0.759	0.730	0.681	0.741
CADD BDA + AMIP all BM	93	0.754	0.779	0.636	0.766
CADD WT + RF + BDA + AMIP all BM	111	0.758	0.736	0.645	0.747

The values shown are the areas under the curve of Receiver Operating Characteristic (ROC) curve analyses
CADD coronary artery disease diagnosis, AMIP acute myocardial infarction prognostication, ACD combined coronary artery disease diagnosis and outcome (AMI) prognostication, SVM support vector machine, DDA diagonal discriminant analysis, LDA linear discriminant analysis, BDA binary discriminant analysis, RF random forests, BM biomarker; ≥ 3, biomarkers present in at least 3 out of the 5 biomarker patterns resulting from the different discovery approaches; ≥ 2, biomarkers present in at least 2 out of the 5 biomarker patterns resulting from the different discovery approaches

ISSN: 1471-2105