npj Biosensing

Table 1 Performance metrics of four classifiers (SVM, LDA, LR, and TREE) evaluated through leave-one-out cross-validation on the PURE dataset.

From: Optimal signal quality index for remote photoplethysmogram sensing

Excellent vs. Unfit
	Index	N_SQI	K_SQI	E_SQI	S_SQI	M_SQI	Z_SQI	R_SQI	P_SQI
Overall	OF₁	0.86	0.79	0.76	0.71	0.63	0.63	0.63	0.6
SVM	SE	0.83	0.85	0.76	0.55	0.26	0.56	0.33	0.6
	PP	0.9	0.78	0.77	0.47	0.41	0.62	0.3	0.42
	F₁	0.86	0.81	0.76	0.76	0.81	0.71	0.75	0.74
LDA	SE	0.87	0.74	0.81	0.56	0.37	0.53	0.46	0.62
	PP	0.87	0.86	0.78	0.46	0.41	0.42	0.61	0.52
	F₁	0.87	0.78	0.79	0.76	0.57	0.7	0.61	0.54
LOGI	SE	0.85	0.79	0.81	0.53	0.37	0.51	0.46	0.62
	PP	0.92	0.79	0.78	0.45	0.41	0.42	0.61	0.53
	F₁	0.88	0.79	0.79	0.73	0.57	0.68	0.61	0.55
TREE	SE	0.84	0.78	0.7	0.58	0.57	0.41	0.54	0.55
	PP	0.81	0.78	0.7	0.58	0.56	0.45	0.51	0.55
	F₁	0.82	0.78	0.7	0.58	0.56	0.43	0.52	0.55

Excellent vs. Acceptable
	Index	N_SQI	E_SQI	K_SQI	S_SQI	M_SQI	Z_SQI	P_SQI	R_SQI
Overall	OF₁	0.78	0.74	0.74	0.72	0.71	0.7	0.66	0.64
SVM	SE	0.86	0.86	0.98	0.91	0.65	0.64	0.73	0.69
	PP	0.73	0.7	0.7	0.62	0.45	0.45	0.57	0.51
	F₁	0.82	0.77	0.77	0.73	0.65	0.75	0.63	0.58
LDA	SE	0.97	0.86	0.94	0.88	0.57	0.82	0.8	0.76
	PP	0.71	0.65	0.67	0.68	0.45	0.64	0.63	0.63
	F₁	0.82	0.74	0.78	0.76	0.76	0.71	0.7	0.69
LOGI	SE	0.93	0.86	0.94	0.86	0.57	0.79	0.8	0.76
	PP	0.72	0.69	0.67	0.66	0.45	0.61	0.63	0.6
	F₁	0.8	0.76	0.78	0.74	0.75	0.69	0.7	0.67
TREE	SE	0.65	0.72	0.6	0.61	0.68	0.7	0.65	0.62
	PP	0.7	0.7	0.63	0.64	0.7	0.69	0.62	0.6
	F₁	0.67	0.71	0.61	0.62	0.69	0.66	0.6	0.61

Acceptable vs. Unfit
	Index	K_SQI	S_SQI	N_SQI	Z_SQI	E_SQI	M_SQI	R_SQI	P_SQI
Overall	OF₁	0.81	0.69	0.67	0.66	0.61	0.53	0.48	0.41
SVM	SE	0.3	0.26	0.23	0.28	0.25	0.3	0.12	0.11
	PP	0.85	0.79	0.68	0.84	0.67	0.55	0.3	0.14
	F₁	0.87	0.68	0.79	0.84	0.7	0.68	0.33	0.29
LDA	SE	0.3	0.26	0.23	0.43	0.19	0.31	0.21	0.26
	PP	0.85	0.39	0.34	0.93	0.31	0.54	0.26	0.51
	F₁	0.87	0.68	0.79	0.72	0.59	0.47	0.57	0.42
LOGI	SE	0.3	0.26	0.23	0.35	0.23	0.31	0.21	0.26
	PP	0.85	0.39	0.33	0.92	0.33	0.79	0.26	0.5
	F₁	0.87	0.79	0.67	0.59	0.67	0.49	0.57	0.41
TREE	SE	0.6	0.4	0.42	0.46	0.3	0.32	0.28	0.33
	PP	0.62	0.42	0.38	0.56	0.3	0.31	0.28	0.32
	F₁	0.61	0.6	0.41	0.49	0.45	0.47	0.42	0.49

Binary classification was performed on rPPG signals processed using CHROM, GREEN, and OMIT methods across three signal quality classes: Excellent, Acceptable, and Unfit. Metrics include Sensitivity (SE), Positive Predictivity (PP), and F₁ score (F₁). The N_SQI index showed superior performance in Excellent vs. Unfit and Excellent vs. Acceptable comparisons, while K_SQI excelled in the Acceptable vs. Unfit comparison. Note: OF₁ represents the overall F₁ score, calculated as the mean of the F₁ scores from the four classifiers.
Bolded numbers indicate the highest classification rate achieved.

Back to article page

Search

Advanced search

Quick links