TABLE 2.

Classification performance of the proposed machine learning method. Note, NT, PHT, and HT denote normotension, prehypertension, and hypertension, respectively. SP stands for specificity, SN stands for sensitivity, ACC stands for accuracy, PRE stands for precision, MCC stands for Matthew’s correlation coefficient, Kappa stands for Cohen’s kappa coefficient, AUC stands for Area under curve. Values in Bold indicate highest scores achieved for each classification per evaluation metric.

Model	Trail	SP (%)	SN (%)	ACC (%)	PRE (%)	F1 score (%)	AUC (%)	MCC (%)	Kappa (%)	Time (s)
Decision tree	NT vs. PHT	84.44	70.88	72.83	72.89	72.85	72.65	45.22	45.22	0.23
	NT vs. HT	80.39	79.60	79.93	80.30	80.02	79.99	59.49	59.34	0.28
	(NT + PHT) vs. HT	96.26	65.88	80.41	80.41	80.41	76.07	52.14	52.14	0.57
AdaBoost	NT vs. PHT	78.74	71.58	75.51	75.47	75.48	75.16	50.44	50.43	1.64
	NT vs. HT	88.22	75.29	82.75	82.73	82.64	81.76	64.44	64.25	1.59
	(NT + PHT) vs. HT	92.26	60.39	83.11	82.56	82.48	76.33	56.73	56.06	2.47
GBDT	NT vs. PHT	83.62	77.54	80.88	80.86	80.86	80.58	61.32	61.30	7.24
	NT vs. HT	93.10	82.35	88.58	88.62	88.48	87.73	76.50	76.30	6.79
	(NT + PHT) vs. HT	96.37	66.27	87.72	87.75	87.15	81.32	68.85	67.63	11.30
Random forest	NT vs. PHT	84.20	76.14	80.57	80.54	80.52	80.17	60.64	60.59	1.30
	NT vs. HT	93.68	83.14	89.22	89.30	89.15	88.41	77.87	77.67	1.48
	(NT + PHT) vs. HT	96.84	61.57	86.71	86.93	85.90	79.20	66.15	64.28	2.49
XgBoost	NT vs. PHT	85.34	78.95	82.46	82.44	82.44	82.15	64.50	64.48	1.34
	NT vs. HT	93.68	87.84	91.21	91.21	91.19	90.76	81.95	81.91	1.09
	(NT + PHT) vs. HT	95.73	74.90	89.75	89.63	89.50	85.32	74.26	73.84	2.02
LightGBM	NT vs. PHT	85.92	80.70	83.57	83.55	83.55	83.31	66.76	66.75	0.87
	NT vs. HT	94.25	87.84	91.54	91.55	91.51	91.05	82.63	82.57	0.81
	(NT + PHT) vs. HT	96.27	74.12	89.98	89.92	89.68	85.24	74.80	74.22	1.13