Skip to main content
. 2020 Jan 21;5(1):e00774-19. doi: 10.1128/mSystems.00774-19

TABLE 3.

Application of neural network model to single species data and merged species dataa

Antimicrobial AUC for the following species (no. of isolates):
E. coli (1,694)
M. tuberculosis (3,528)
S. enterica (658)
S. aureus (1,236)
E. coli, M. tuberculosis, S. enterica, and S. aureus
Validation data Test data Validation data Test data Validation data Test data Validation data Test data Validation data Test data
Rifampin NA NA 0.94 ± 0.04 0.94 NA NA NA NA 0.94 ± 0.04 0.95
Isoniazid NA NA 0.88 ± 0.04 0.94 NA NA NA NA 0.93 ± 0.03 0.94
Streptomycin NA NA 0.89 ± 0.06 0.87 NA NA NA NA 0.89 ± 0.06 0.89
Ethambutol NA NA 0.93 ± 0.06 0.87 NA NA NA NA 0.91 ± 0.05 0.92
Pyrazinamide NA NA 0.82 ± 0.11 0.83 NA NA NA NA 0.85 ± 0.04 0.87
Ciprofloxacin 0.99 ± 0.01 0.97 0.90 ± 0.12 0.97 0.75 ± 0.15 0.85 0.98 ± 0.01 0.99 0.98 ± 0.01 0.97
a

The results for M. tuberculosis were slightly different from the results shown in Table 2, as the output dimension was changed. NA, the result is not available.