Predicting Increased Blood Pressure Using Machine Learning
Table 2
Predictors, deviance, misclassification, and pseudo- by tree.
Tree
Predictors
Women
Men
Deviance
Misclassification error rate
Pseudo-
Deviance
Misclassification error rate
Pseudo-
1
BMI
104.50
0.27
0.32
89.35
0.26
0.16
2
WC
149.30
0.40
0.03
92.96
0.27
0.13
3
HC
148.90
0.41
0.03
89.59
0.26
0.16
4
WHR
139.50
0.33
0.09
95.09
0.29
0.11
5
BMI + WC
109.50
0.26
0.29
71.07
0.21
0.33
6
BMI + HC
108.10
0.26
0.29
80.17
0.25
0.25
7
BMI + WHR
104.60
0.23
0.32
79.9
0.23
0.25
8
WC + HC
149.30*
0.40
0.03
82.19
0.23
0.23
9
WC + WHR
115.90
0.26
0.24
69.54
0.24
0.35
10
HC + WHR
118.60
0.29
0.23
76
0.21
0.29
11
BMI + WC + HC
94.24
0.22
0.39
72.66
0.23
0.32
12
WC + HC + WHR
99.17
0.23
0.35
61.5
0.17
0.42
13
BMI + WC + WHR
87.42
0.19
0.43
64.98
0.19
0.39
14
BMI + HC + WHR
101.10
0.22
0.34
61.93
0.16
0.42
15
BMI+ WC + HC + WHR
89.46
0.19
0.42
57.25
0.16
0.46
HC dropped out. Notice that the endpoint for women is systolic blood pressure greater than 120 mmHg (prehypertension), while for men is greater than 140 mmHg (hypertension).
