Skip to main content

Table 2 DICs, log-likelihood, and posterior parameters in each mechanism model

From: Explaining age disparities in tuberculosis burden in Taiwan: a modelling study

Model

Indicators

Posterior parameter distribution

DIC

logL

m0

m1

m2

m3

γ

bbase

σ(C)

σ(E)

bini

bend

t0

w(A)

w(E)

ε

m0

9090

− 4546 (− 4552,-4541)

1.78 (1.33, 2.41)

7.70 (6.09, 9.86)

-

-

-

-

-

-

-

-

Single-mechanism model

m1

3288

− 1643 (− 1651, − 1638)

1.33 (1.32, 1.39)

5.35 (5.25, 5.55)

0.100 (0.100, 0.100)

9.99 (9.93, 10.0)

-

-

-

-

-

-

m2

3099

− 3048 (− 3055, − 3043)

2.96 (2.81, 3.03)

-

-

-

9.90 (9.47, 10.0)

2.02e-3 (4.05e-4, 8.67e-3)

1957.0 (1957.0, 1957.1)

-

-

-

m3

2455

− 1228 (− 1235, − 1223)

7.24 (5.87, 7.65)

15.2 (12.4, 16.4)

-

-

-

-

-

3.54 (3.46, 3.59)

5.00 (4.98, 5.00)

1.01e-2 (1.00e-2, 1.03e-2)

Two-mechanism model

m12

− 1159

437 (432, 440)

1.98 (1.37, 2.54)

-

0.101 (0.100, 0.106)

9.06 (8.25, 9.82)

8.48 (6.29, 9.91)

0.593 (0.203, 1.06)

1986.0 (1977.3, 1989.9)

-

-

-

m13

− 242

118 (113, 122)

3.23 (1.36, 7.50)

10.9 (5.50, 23.7)

0.104 (0.100, 0.119)

1.02 (1.00, 1.09)

-

-

-

1.57 (1.46, 1.80)

4.80 (4.23, 4.99)

0.165 (0.127, 0.201)

m23

1012

− 504 (− 512, − 498)

7.39 (6.45, 7.66)

-

-

-

28.2 (24.8, 30.0)

4.64 (3.93, 5.23)

1957.2 (1957.0, 1957.8)

2.06 (1.98, 2.13)

5.00 (4.99, 5.00)

1.00e-2 (1.00e-2, 1.02e-2)

Three-mechanism model (‘full’ model)

m123

− 5825

462 (458, 464)

5.94 (2.83, 7.58)

-

0.137 (0.103, 0.209)

5.50 (3.79, 7.62)

8.99 (4.34, 13.7)

4.02 (1.84, 6.55)

1999.9 (1998.3, 2001.5)

1.76 (1.25, 2.70)

1.49 (1.10, 2.10)

0.217 (0.035, 0.455)

  1. The log-likelihood, and central estimates and 95% credible intervals (in brackets) of the key posterior parameters are presented for each mechanism model. For comparison across models, infection rates bbase, bini, and bend reported in this table have been standardised by scaling the age-specific ‘mixing’ matrix so that the mean of its elements equals to 1. Details of parameters and mechanism models are described in the ‘Method’ section and Table 1. Abbreviations: DIC - deviance information criterion; logL – log-likelihood