BioMed Research International / 2019 / Article / Tab 5

Research Article

Comparison of Two Bayesian Methods in Evaluation of the Absence of the Gold Standard Diagnostic Tests

Table 5

The posterior estimation of four models using the TB data under conditional independence situation.

SituationMethodModelKnotMeanSDMedian95% Bayesian CI
(P2.5-P97.5)

No prior constraintsBayesian probabilistic constraint modelNP0.6600.2430.7620.503-0.841
0.4180.2510.4260.203-0.584
0.5110.2510.5680.302-0.688
0.5690.2490.6310.377-0.759
0.5120.1940.5150.361-0.667
Conditional
Covariance
Bayesian model
NC0.6180.2500.6310.071-0.974
0.3770.2500.2950.025-0.929
0.4680.2550.4350.036-0.947
0.5260.2530.4800.055-0.962
0.4980.1940.4970.148-0.850

Prior constraintsBayesian probabilistic constraint modelPP0.7380.0360.7390.714-0.762
0.4540.0510.4530.419-0.487
0.5850.0310.5850.563-0.606
0.5250.0280.5250.506-0.544
0.5340.0420.5340.506-0.562
Conditional
Covariance
Bayesian model
PC0.8980.0390.9100.814-0.963
0.7650.1250.7750.515-0.968
0.5940.0480.5860.523-0.712
0.6790.0480.6760.592-0.780
0.6360.0860.6500.435-0.773

: sensitivity of T-SPOT; : specificity of T-SPOT; : sensitivity of KD38; : specificity of KD38;: prevalence within the patients in the study; SD: standard deviation; CI: confidence interval.