A New Model Selection Metric for Biomarker Detection Algorithms and Tools
Table 4
Simulation result when Pc = 0.05.
Cutoff
Sen
Spe
PPV
NPV
St
N_random
N_screening
L
Cost
F
Gmeans
7
0.996
0.909
0.367
1
0.218
330.835
6642.901
0.633
99512490
8.979
0.952
8
0.99
0.952
0.525
0.999
0.251
249.6
5041.961
0.476
75379568
8.695
0.971
9
0.977
0.977
0.697
0.999
0.288
190.419
3897.606
0.304
58017996
8.336
0.977
10
0.952
0.99
0.84
0.997
0.319
155.481
3266.492
0.163
48213020
7.918
0.971
11
0.909
0.996
0.928
0.995
0.338
138.375
3046.039
0.077
44297910
7.453
0.951
12
0.841
0.999
0.972
0.992
0.347
130.818
3111.103
0.037
44192872
7.012
0.917
13
0.747
1
0.99
0.987
0.351
127.641
3418.326
0.023
46947320
6.861
0.864
14
0.63
1
0.997
0.981
0.353
126.581
4019.678
0.023
52854922
6.965
0.794
15
0.5
1
0.999
0.974
0.353
126.196
5054.243
0.027
63161986
7.163
0.707
Note: Sen: sensitivity; Spe: specificity; St: standardized treatment effect; N_random: randomized sample size; N_screening: screening sample size; L: predictability loss of M; cost: estimated total cost in an M-based drug clinical trial.
