Gene expression biclustering analysis is a commonly used technique to see the interaction between genes under certain experiments or conditions. More specifically in the study of diseases, these methods are used to compare control and affected data in order to identify the involved or relevant genes. In some cases, discretization is needed for these algorithms to work correctly. In this context, the choice of the discretization method is extremely important and has a major impact on the outcome. In this work we analyze several discretization methods for Alzheimer Disease (AD) gene expression data and compare the results of a state-of-art biclustering algorithm after each discretization. The comparison reveals that biclusters obtained from discretized expression values achieve a major coverage and overall enrichment than biclusters generated from real-valued expression data. In a particular experiment, a clustering-based discretization method overcomes all competing techniques for the dataset under study, in statistical terms.