Does latent classification analysis need a separate verification bias removal? how to perform that?

Question

[sukeshinik5]

I have clinical data which lacks a golden standard for diagnosis. instead, multiple indices are used with different accuracy. I have transformed the results from these indices to binary for patients having and not having the disease. This is then used for latent classification analysis. I want to know if I should use all samples for this analysis. or should I divide them and then use the remaining samples for validation? Also, will I need to check for verification bias and how do I select a method to do that? How to check the category of each patient for diagnosis after that?

A tabular form with binary data of 1 and 2 for disease present and absent is used to subject under latent classification analysis (LCA) using the poLCA package in R. The results I get are as follows

Fit for 2 latent classes: number of observations: 65 number of estimated parameters: 7 residual degrees of freedom: 0 maximum log-likelihood: -116.6411

AIC(2): 247.2821 BIC(2): 262.5028 G^2(2): 0.3041395 (Likelihood ratio/deviance statistic) X^2(2): 0.291966 (Chi-square goodness of fit)

What type of graph will make me understand the results better? Thank you, your help will be appreciated here

[sukeshinik5]

I have not provided codes. Please find them below.

#create a latent class with log-linear independence and 3 classes of data.

    f = cbind(index1, index2, index3)~1

    nes2<-poLCA(f, data1, nclass = 2, nrep= 10, verbose = FALSE, graphs = TRUE)

 ##nclass can be decided by the number of classes you want

    install.packages("sBIC")

    library(sBIC)

#run a sequence of models with 1-10 classes and print out the model with the lowest BIC

    max_II <- -100000

    min_bic <- 100000

    for(i in 2:10) {


    lc <- poLCA(f, data1, nclass=i, maxiter=3000, 
              tol=1e-5, na.rm=FALSE,  
              nrep=10, verbose=TRUE, calc.se=TRUE)

    if(lc$bic < min_bic){
    min_bic <- lc$bic
    LCA_best_model<-lc }
}       

    LCA_best_model