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

Table 3

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

Classifier	TPR (%)	TNR (%)	PPV (%)	NPV (%)	TA (%)	AUC	LB	UP
Decision tree (DT)
Fine	81	68	68	81	75.9	0.77	0.19	0.68
Coarse	86	64	74	79	77.6	0.80	0.14	0.64
Support vector machine (SVM)
Linear	99	55	96	78	81.9	0.80	0.01	0.55
Quadratic	92	64	82	80	81.0	0.84	0.08	0.64
Cubic	85	61	71	78	75.9	0.78	0.15	0.61
Med. Gaussian	90	52	77	76	75.9	0.81	0.10	0.52
K-nearest neighbor (KNN)
Fine	78	55	60	74	69.0	0.66	0.22	0.56
Medium	89	43	70	72	71.6	0.78	0.11	0.43
Cosine	89	48	72	74	73.3	0.78	0.11	0.48
Ensemble classifiers
Bagged tree	85	66	73	80	77.6	0.81	0.15	0.66
Subsp. disc.	99	34	94	71	74.1	0.77	0.01	0.34
RUSBoosted tree	85	66	73	80	77.6	0.79	0.15	0.66