Exploratory study on classification of chronic obstructive pulmonary disease combining multi-stage feature fusion and machine learning

Table 3 Performance evaluation of the MSFF on the stage data D1–D4 generated by cross-validation method

Phased data set	Model	acc	sn	sp	mcc	Curve
D1	RF	0.750	0.773	0.722	0.495	0.748
	SVM	0.725	0.864	0.556	0.445	0.710
	KNN	0.675	1.000	0.278	0.418	0.639
	PLS	0.775	0.773	0.778	0.548	0.775
D2	RF	0.750	0.773	0.722	0.495	0.748
	SVM	0.700	0.864	0.500	0.395	0.682
	KNN	0.600	0.773	0.389	0.175	0.581
	PLS	0.750	0.818	0.667	0.492	0.742
D3	RF	0.800	0.909	0.667	0.601	0.788
	SVM	0.775	0.818	0.722	0.544	0.770
	KNN	0.625	0.773	0.444	0.231	0.609
	PLS	0.725	0.818	0.611	0.441	0.715
D4	RF	0.825	0.864	0.778	0.646	0.821
	SVM	0.800	0.864	0.722	0.595	0.793
	KNN	0.700	0.909	0.444	0.406	0.677
	PLS	0.850	0.864	0.833	0.697	0.849

D1, D2, D3 and D4 represent the AECOPD patient data of stage 1 to 4. At each stage, there are 408 samples

ISSN: 1472-6947