[Feature Request]: Comparison of different machine learning tests using logit of difference (AUC)

[laislvsantana]

Description

Comparison of two diagnostic models

Purpose

A function that allows you to perform a statistical analysis of two test models from the following data: comparison title [Algorithm X vs Algorithm Y], logit difference, sample size

Use-case

Compare the performance of different tests

Is your feature request related to a problem?

Not all machine learning studies provide an AUC (c-statistic) with standard error and confidence intervals. In these cases, it is not feasible to use the data. However, there is an alternative: it is possible to estimate the AUC from the sensitivity and specificity and then convert to the logit function. The comparison of the two testicles is then made by the logit difference (AUC).

Is your feature request related to a JASP module?

Meta Analysis

Describe the solution you would like

A function that would allow us to compare the performance of machine learning algorithms, providing the logit difference.

Describe alternatives that you have considered

This article provided the R code used to carry out the aforementioned statistical analysis (comparison of different tests, using logit difference (AUC))

Additional context

Some meta-analyses that made use of this analysis:

Article: https://pubmed.ncbi.nlm.nih.gov/33200995/

Article: https://pubmed.ncbi.nlm.nih.gov/30763612/

[FBartos]

I'm currently too busy working on other projects--should be reassigned to someone else.

[JorisGoosen]

@sophieberkhout ?

[sophieberkhout]

Sorry, I also do not have time for this. I am also not sure whether this should be in the meta-analysis module. While the outcome of the requested analysis (AUC) could be used in a meta-analysis, the analysis itself is more machine learning related I think.