Fig 1.
Applied SEIR model [15].
Table 1.
Comparison of the cell characteristics between the in-sample and out-of-sample.
Fig 2.
Distribution of the PTB IR based on the TB transmission probability estimated using the five dynamic models.
The green scatter plots indicate the observed data for individual prison cells. The dark blue line indicates the fitted line. The two vertical dashed lines represent the fitted data interval.
Table 2.
Fitted models of the negative binomial regression (NBRG) for the five dynamic models (n = 652).
Table 3.
In-sample goodness-of-fit and discrimination test results of five dynamic models.
Fig 3.
The 10-fold cross-validation results of four dynamic models compared to the Wells–Riley model.
Light blue represents Rudnick & Milton (ACH), blue—Rudnick & Milton (L/s/p) model, yellow—Issarow et al.’s model, and red—Applied SEIR model.
Fig 4.
Calibration plots depicting the external validation of the five dynamic models.
Table 4.
Comparison of accuracy and bias metrics from external validation (out-of-sample, n = 333).