Tree-Based Risk Factor Identification and Stroke Level Prediction in Stroke Cohort Study

Table 1

Description of the data.

Factor	2016			2017			2018
	No.	Ratio			Ratio			Ratio
All	4296			3915			4180
Stroke levels ()
TIA: 0	88	2%		88	2.2%		100	2.4%
Low risk: 1	3174	73.9%		2916	74.5%		2920	69.9%
Medium risk: 2	446	10.4%		433	11.1%		491	11.7%
High risk: 3	553	12.9%		442	11.3%		622	14.9%
Stroke: 4	35	0.8%		36	0.9%		47	1.1%
Continuous variable
Age ()	4296		<0.01	3915		<0.01	4180		<0.01
	Mean	std		Mean	std		Mean	std
TIA: 0	69.96	(12.34)		66.45	(11.23)		66.22	(11.31)
Low risk: 1	60.39	(11.41)		56.94	(11.36)		58.10	(11.40)
Medium risk:2	65.15	(11.75)		61.52	(11.73)		62.09	(11.66)
High risk: 3	66.58	(10.30)		64.19	(9.93)		63.86	(10.26)
Stroke: 4	68.00	(9.03)		66.17	(8.50)		64.98	(9.27)
Discrete variables
Gender ()			0.001			0.011			0.115
Female: 0	2240	52.1%		2013	51.4%		2138	51.1%
Male: 1	2056	47.9%		1902	48.6%		2042	48.9%
History of stroke ()			<0.01			<0.01			<0.01
Yes: 1	35	0.8%		36	0.9%		47	1.1%
No: 0	4261	99.2%		3879	99.1%		4133	98.9%
History of TIA ()			<0.01			<0.01			<0.01
Yes: 1	92	2.1%		91	2.3%		104	2.5%
No: 0	4204	97.9%		3824	97.7%		4076	97.5%
Family history of stroke ()			<0.01			<0.01			<0.01
Yes: 1	195	4.5%		177	4.5%		223	5.3%
No: 0	4101	95.5%		3738	95.5%		3957	94.7%
AF or VHD ()			<0.01			<0.01			<0.01
Yes: 1	112	2.6%		101	2.6%		119	2.8%
No: 0	4184	97.4%		3814	97.4%		4061	97.2%
Hypertension ()			<0.01			<0.01			<0.01
Yes: 1	684	15.9%		656	16.8%		938	22.4%
No: 0	3612	84.1%		3259	83.2%		3242	77.6%
Dyslipidemia ()			<0.01			<0.01			<0.01
Yes: 1	943	22%		1167	29.8%		1740	41.6%
No: 0	3353	78%		2748	70.2%		2440	58.4%
Diabetes ()			<0.01			<0.01			<0.01
Yes: 1	480	11.2%		409	10.4%		495	11.8%
No: 0	3816	88.8%		3506	89.6%		3685	88.2%
Smoking history ()			<0.01			<0.01			<0.01
Yes: 1	414	9.6%		291	7.4%		353	8.4%
No: 0	3882	90.4%		3624	92.6%		3827	91.6%
AO or obesity ()			<0.01			<0.01			<0.01
Yes: 1	679	15.8%		222	5.7%		245	5.9%
No: 0	3617	84.2%		3693	94.3%		3935	94.1%
Lack of exercise ()			<0.01			<0.01			<0.01
Yes: 1	524	12.2%		444	11.3%		556	13.3%
No: 0	3772	87.8%		3471	88.7%		3624	86.7%

TIA: transient ischemic attack; AF: atrial fibrillation; VHD: valvular heart disease; AO: apparently overweight. Significance analyses were performed by analysis of variance. All tests were two-sided. Statistically significant values ( ); statistically very significant values ().