Table 7.
Performance of one-tiered (A) and two-tiered (B) classification strategies using SVMa
| Model | Strategy | Classification Rate | Overall Accuracy | ||||||
|---|---|---|---|---|---|---|---|---|---|
| 1A (Strong) | 1B (Weak) | Neg (Nonsensitizers) | |||||||
| Correct | Under | Over | Correct | Under | Correct | Over | |||
| LLNA | A | 83 ± 12% (29/35) |
17 ± 12% (6/35) |
17 ± 10% (9/52) |
69 ± 12% (36/52) |
14 ± 9% (7/52) |
85 ± 12% (28/33) |
15 ± 12% (5/33) |
78 ± 7% (93/120) |
| B | 91 ± 9% (32/35) |
9 ± 9% (3/35) |
15 ± 10% (8/52) |
81 ± 11% (42/52) |
4 ± 5% (2/52) |
97 ± 6% (32/33) |
3 ± 6% (1/33) |
88 ± 6% (106/120) |
|
| Human | A | 85 ± 14% (22/26) |
15 ± 14% (4/26) |
26 ± 15% (8/31) |
61 ± 17% (19/31) |
13 ± 12% (4/31) |
80 ± 14% (24/30) |
20 ± 14% (6/30) |
75 ± 9% (65/87) |
| B | 85 ± 14% (22/26) |
15 ± 14% (4/26) |
22 ± 15% (7/31) |
65 ± 17% (20/31) |
13 ± 12% (4/31) |
93 ± 9% (28/30) |
7 ± 9% (2/30) |
81 ± 8% (70/87) |
|
LLNA, murine local lymph node assay; SVM, support vector machine.
The values after ± indicate 95% confidence limits of proportion for correct classification rate.
a
Leave-one-out cross-validation results.