Detecting Congestive Heart Failure by Extracting Multimodal Features and Employing Machine Learning Techniques

Table 2

CHF detection performance based on frequency domain features by applying machine learning techniques.

Classifier	TPR (%)	TNR (%)	PPV (%)	NPV (%)	TA (%)	AUC	LB	UP
Decision tree (DT)
Fine	83	75	73	85	80.2	0.84	0.17	0.75
Coarse	93	64	85	81	81.9	0.81	0.07	0.64
Support vector machine (SVM)
Linear	82	77	72	86	80.2	0.86	0.18	0.77
Quadratic	82	82	73	88	81.9	0.88	0.18	0.82
Cubic	88	68	77	82	80.2	0.83	0.13	0.58
Med. Gaussian	90	77	83	87	85.3	0.90	0.16	0.77
K-nearest neighbor (KNN)
Fine	85	75	75	85	81.0	0.86	0.15	0.75
Medium	89	66	78	81	80.2	0.88	0.11	0.66
Cosine	64	73	55	79	67.2	0.75	0.36	0.73
Ensemble classifiers
Bagged tree	85	77	76	86	81.9	0.88	0.15	0.77
Subsp. disc.	86	70	76	83	80.2	0.85	0.14	0.70
RUSBoosted tree	79	75	69	84	77.6	0.81	0.21	0.75