. Author manuscript; available in PMC: 2009 Jul 24.

Published in final edited form as: Artif Intell Med. 2009 Jan 21;46(3):217–231. doi: 10.1016/j.artmed.2008.12.004

Table 3.

Decision rule sets for DT with MLP selected features (a) full, (b) pruned (level 1) and (c) pruned (level 2)

Rule #	If	and	and	and	and	then	Class	Membership
	DBP_avg (mm Hg)	DBP_krt	pCO_2max (mm Hg)	SBP_max (mm Hg)	DBP_avg (mm Hg)		Y (PVL = 1)/N (PVL = 0)	Class/total	%
(a) Full
1	<42	>2.7					Y	8/9	90
2	<42	<2.7	>58				Y	4/4	100
3	<42	<2.7	<58	<73			Y	6/7	86
4	<42	<2.7	<58	<73	<30		Y	2/2	100
5	<42	<2.7	<58	<73	>30		N	14/18	78
6	>42	<-0.5					N	9/9	100
7	>42	<1.9					Y	6/9	67
		>-0.5
8	>42	>1.9					N	11/11	100
Rule #	If	and	and	and	and	then	Class	Membership
	DBP_avg (mm Hg)	DBP_krt	pCO_2max (mm Hg)	SBP_max (mm Hg)	DBP_avg (mm Hg)		Y (PVL = 1)/N (PVL = 0)	Class/total	%

(b) Pruned (level 1)
1	<42	>2.7					Y	8/9	90
2	<42	<2.7	>58				Y	4/4	100
3	<42	<2.7	<58	<73			Y	6/7	86
4	<42	<2.7	<58	<73	<30		Y	2/2	100
5	<42	<2.7	<58	<73	>30		N	14/18	78
6	>42						N	23/29	79
Rule #	If	and	and	and	and	then	Class	Membership
	DBP_avg (mm Hg)	DBP_krt	pCO_2max (mm Hg)	SBP_max (mm Hg)	DBP_avg (mm Hg)		Y (PVL = 1)/N (PVL = 0)	Class/total	%

(c) Pruned (level 2)
1	<42	>2.7					Y	8/9	90
2	<42	<2.7	>58				Y	4/4	100
3	<42	<2.7	<58	<73			Y	6/7	86
4	<42	<2.7	<58	>73			N	14/20	70
5	>42						N	23/29	79