Таблица 1. Статистические параметры бинарной классификации BBB+/BBB- лекарств методом логистической регрессии.
Sets |
N |
Descriptors |
TP |
FN |
TN |
FP |
Sens |
Spec |
Acc |
MCC |
training CV test |
67
16 |
AlogP |
32 32 6 |
8 8 4 |
14 14 5 |
13 13 1 |
0.800 0.800 0.600 |
0.519 0.519 0.833 |
0.687 0.687 0.688 |
0.333 0.333 0.423 |
training CV test |
67
16 |
AlogP, HBD |
34 34 5 |
6 6 5 |
16 12 6 |
11 12 0 |
0.850 0.850 0.500 |
0.593 0.556 1.000 |
0.748 0.731 0.688 |
0.462 0.429 0.522 |
training CV test |
67
16 |
AlogP, HBD, MW |
31 31 5 |
9 9 5 |
17 15 6 |
10 12 0 |
0.775 0.775 0.500 |
0.630 0.556 1.000 |
0.716 0.687 0.688 |
0.407 0.338 0.522 |
training CV test |
67
16 |
AlogP, HBD, TPSA |
34 34 6 |
6 6 4 |
20 20 6 |
7 7 0 |
0.850 0.850 0.600 |
0.741 0.741 1.000 |
0.806 0.806 0.750 |
0.595 0.595 0.600 |
training CV test |
67
16 |
AlogP, HBD, MW, TPSA |
33 32 6 |
7 8 4 |
20 19 6 |
7 8 0 |
0.825 0.800 0.600 |
0.741 0.704 1.000 |
0.791 0.761 0.750 |
0.566 0.504 0.600 |
training CV test |
67
16 |
α, ∑Ca, ∑Cd |
35 32 7 |
5 8 3 |
18 17 6 |
9 10 0 |
0.875 0.800 0.700 |
0.667 0.630 1.000 |
0.791 0.731 0.812 |
0.560 0.436 0.683 |