Predicting blood pressure from physiological index data using the SVR algorithm

BMC Bioinformatics

Table 6 Comparison of the blood pressure predictions of each model

Index		SVR		LinearR		BP
		Ps	Pd	Ps	Pd	Ps	Pd
Accuracy(%)	±3	96.43	97.14	17.43	34.00	30.86	54.00
	±5	97.71	98.43	30.71	51.14	45.00	73.14
	±7	98.43	99.00	40.57	69.43	56.86	84.00
	±10	99.00	99.43	57.14	87.14	69.43	96.43
Pass rate(%)	≤0.03	96.57	96.00	22.57	22.57	36.71	36.71
	≤0.05	98.00	97.14	38.29	35.71	51.86	56.57
	≤0.07	99.00	97.86	52.14	47.29	65.14	69.43
	≤0.10	99.71	99.00	65.29	63.57	75.71	82.86
MAPE(%)		0.3173	0.4875	8.6916	8.5605	6.6690	5.8319
MAE		0.4135	0.3374	11.1090	5.5047	8.4562	3.7260
R ²		0.9835	0.9665	0.0896	0.0546	0.3857	0.5025
Spearman		0.9917	0.9911	0.3341	0.2696	0.5900	0.7182

±3, ±5, ±7, ±10 and ≤0.03, ≤0.05, ≤0.07, ≤0.1 indicate the difference in error range and relative error range between the actual and predicted values

ISSN: 1471-2105