Meta4diag

Description The PC prior is defined for the standard deviation σ using

P(σ > u)=α,

i.e. the probability of the standard deviation being larger than a value u is α. The σ is exponential distributed with rate λ under the assumption of a contant penalisation rate, where

λ =

log(α) / u

u:
α:

Reset

Check

α:
β:

Reset

Check

Description The half-Cauchy prior is defined for the standard deviation σ using

σ ~

2α / π(α²+σ²),

where α is the scale parameter.

Reset

Check

μ:
τ²:

Reset

Check

Description The inverse-Wishart prior is defined for the covariance matrix Σ, where

Σ ~ Wishart^-1₂ ( ν, R),

and

R =

R₁₁	R₁₂
R₂₁	R₂₂

ν:
R₁₁:
R₂₂:
R₁₂:

Reset

Second Variance

Description The PC prior is defined for the standard deviation σ using

P(σ > u)=α,

i.e. the probability of the standard deviation being larger than a value u is α. The σ is exponential distributed with rate λ under the assumption of a contant penalisation rate, where

λ =

log(α) / u

u:
α:

Reset

Check

α:
β:

Reset

Check

Description The half-Cauchy prior is defined for the standard deviation σ using

σ ~

2α / π(α²+σ²),

where α is the scale parameter.

Reset

Check

μ:
τ²:

Reset

Check

Correlation

Description There are 3 strategies to define a PC prior for the correlation parameter ρ. To see the details, please choose from the following strategy:

Strategy 1 is defined using

P(ρ < ρ₀ | ρ₀) = ω,

P(ρ < u₁ | ρ₀) = α₁,

where ρ₀ is the base model, i.e. one believes that ρ is around ρ₀; u₁ is a value that is smaller than ρ₀; α₁ is the probability of the left tail event.

Strategy 2 is defined using

P(ρ < ρ₀ | ρ₀) = ω,

P(ρ > u₂ | ρ₀) = α₂,

where ρ₀ is the base model, i.e. one believes that ρ is around ρ₀; u₂ is a value that is larger than ρ₀; α₂ is the probability of the right tail event.

Strategy 3 is defined using

P(ρ < u₁ | ρ₀) = α₁,

P(ρ > u₂ | ρ₀) = α₂,

where ρ₀ is the base model, i.e. one believe that ρ is around ρ₀; u₁ is a value that is smaller than ρ₀; α₁ is the probability of the left tail event; u₂ is a value that is larger than ρ₀; α₂ is the probability of the right tail event.

ρ₀:
ω:
u₁:
α₁:

Reset

Check

ρ₀:
ω:
u₂:
α₂:

Reset

Check

ρ₀:
u₁:
α₁:
u₂:
α₂:

Reset

Check

Description The normal prior is defined for the correlation ρ such that

ϑ ~ N(μ, σ²),

where

logit^-1(ϑ) =

ρ+1 / 2

is a reparameterized ρ using Fisher's z-transformation.

μ:
σ²:

Reset

Check

a:
b:

Reset

Check

0.05	0.1
0.15	0.2
0.25	0.75
0.8	0.85
0.9	0.95

Title:
Title Size:
Axis Size:
Axis Name Size:

Left	Right
Bottom	Not show

Sens	Spec
FPR	FNR
LR+	LR-
LLR+	LLR-
DOR	LDOR

Left	Right
Center	Not show

Meta4diag: Bayesian Bivariate Meta-analysis of Diagnostic Test Studies

Welcome to meta4diag world!

True Positive(TP)	False Positive(FP)
False Negative(FN)	True Negative(TN)

Meta4diag: Bayesian Bivariate Meta-analysis of Diagnostic Test Studies

Welcome to meta4diag world!

SROC Lines:

Tutorial:

First Variance PC Prior Plot:

First Variance Inverse-Gamma Prior Plot:

First Variance Half-Cauchy Prior Plot:

First Variance Truncated-Normal Prior Plot:

First Variance Uniform Prior Plot:

Second Variance PC Prior Plot:

Second Variance Inverse-Gamma Prior Plot:

Second Variance Half-Cauchy Prior Plot:

Second Variance Truncated-Normal Prior Plot:

Second Variance Uniform Prior Plot:

Correlation PC Prior Strategy 1 Plot:

Correlation PC Prior Strategy 2 Plot:

Correlation PC Prior Strategy 3 Plot:

Correlation Normal Prior Plot:

Correlation Beta Prior Plot:

Correlation Uniform Prior Plot: