Please Select a model

Imperfect Gold Standard Models

Conventionally, we estimate accuracy of a new diagnostic test by comparing with a gold standard. This requires the assumption that accuracy of the gold standard is perfect (100% sensitivity and 100% specificity). However, if the true accuracy of the gold standard is imperfect or unknown, estimated accuracy of the new diagnostic test is likely to be inaccurate. Bayesian latent class models (LCMs) can be helpful under these circumstances, but the necessary analysis requires expertise in computational programming.

Here, we developed open-access web-based applications that allow non-experts to apply Bayesian LCMs to their own data sets via a user-friendly interface.

The 2-tests in 2-populations Model

(Simplified Interface)

**History**

In 1980, Hui and Walter proposed the first statistical model to estimate the accuracy of diagnostic tests when the accuracy of the gold standard is unknown. Their model does not assume that the gold standard is perfect, but calculates the accuracy of diagnostic tests based on the estimation of true disease prevalence. Their approach requires that two diagnostic tests are both applied to two populations with differing disease prevalence. Disease prevalence in both populations and the accuracy of both diagnostic tests can then be estimated using the formula.

**What we provide **

Here you can run Hui and Walter model (2-tests in 2-populations model) within a click. You are not required to use formula, or download any programs to your computer.

**What you need**

A data set in a summary table format.

More ..The 2-tests in 2-populations Model

(Advanced Interface)

**History**

In 1980, Hui and Walter proposed the first statistical model to estimate the accuracy of diagnostic tests when the accuracy of the gold standard is unknown. Their model does not assume that the gold standard is perfect, but calculates the accuracy of diagnostic tests based on the estimation of true disease prevalence. Their approach requires that two diagnostic tests are both applied to two populations with differing disease prevalence. Disease prevalence in both populations and the accuracy of both diagnostic tests can then be estimated using the formula.

**What we provide **

Here you can run Hui and Walter model (2-tests in 2-populations model) within a click. You are not required to use formula, or download any programs to your computer. **With this advanced interfaces, experienced researchers can modify all settings in the models as needed. **

More ..The 3-tests in 1-population Model

(Simplified Interface)

**History**

Based on the same principle of Hui and Walter model, in 1988 Walter and Irwig expanded the model for the application of three diagnostic tests in one population. Disease prevalence and the true accuracy of all three diagnostic tests can then be estimated using the maximum likelihood estimation methods. None of the commonly used statistical softwares such as SAS, SPSS and STATA contain the commands for this model.

**What we provide **

Here you can run Walter and Irwig model (3-tests in 1-population model) within a click.

**What you need**

A data set in a summary table format.

More ..The 3-tests in 1-population Model

(Advanced Interface)

**History**

Based on the same principle of Hui and Walter model, in 1988 Walter and Irwig expanded the model for the application of three diagnostic tests in one population. Disease prevalence and the true accuracy of all three diagnostic tests can then be estimated using the maximum likelihood estimation methods. None of the commonly used statistical softwares such as SAS, SPSS and STATA contain the commands for this model.

**What we provide **

Here you can run Walter and Irwig model (3-tests in 1-population model) within a click. **With this advanced interfaces, experienced researchers can modify all settings in the models as needed. **

More ..4-tests in 1-population Model (Simplified Interface)

**History**

In 1995, Joseph et al proposed Bayesian latent class model (LCM) to estimate disease prevalence and accuracies of diagnostic tests in the imperfect gold standard model developed by Hui and Walter. The 4-tests in 1-population model was described and analyzed using Bayesian LCM. However, Bayesian estimation requires specialized statistical software such as R and winBUGS.

**What we provide **

Here we provide 4-tests in 1-population model, which can be analyzed with a simple click.

**What you need**

A data set in a summary table format.

More ..4-tests in 1-population Model (Advance Interface)

**History**

In 1995, Joseph et al proposed Bayesian latent class model (LCM) to estimate disease prevalence and accuracies of diagnostic tests in the imperfect gold standard model developed by Hui and Walter. The 4-tests in 1-population model was described and analyzed using Bayesian LCM. However, Bayesian estimation requires specialized statistical software such as R and winBUGS.

**What we provide**

Here you can adjust settings of 4-tests in 1-population mode. For instance, you can add correlation between related tests, adjust shape of prior distribution, input initial values and much more.

More ..Models of Malaria Infections

All models for malaria infections were developed by Mathematical modelling group, Mahidol-Oxford Tropical Medicine Research Unit. If you have any specific questions about these models, please feel free to contact

Sompob Saralamba

**NOTE:** To view and run models in this section, you must have

Wolfram CDF player or

Mathematica installed on your computer.

A Model of Plasmodium Falciparum Population Dynamics in a Patient during Treatment with Artesunate

This is a model that looks at the dynamics of Plasmodium Falciparum population in a patient during treatment with Artesunate

To run the model please download Plugin at

http://www.wolfram.com/cdf-player

For more information or details regarding the model, please contact

Sompob Saralamba

Download and Run ..