The grand challenge of Alzheimer's disease (AD) are the early detection, accurately diagnoising, and the the reconstruction of genes signal pathways and its regulatory network. In our study, we use informatics methods to extract significant genes and reconstruct AD gene regulatory network. To avoid the limitation of traditional clustering methods which group genes in only one class and base on the global similarities in their expression profiles, we provide a data-driven biclustering method, independent component analysis (ICA), to identify significant genes and gene regulatory modules in a meta-analysis of gene expression data of Alzheimer's disease (AD). According to the function of brain area, we use the gene expression data of normal and AD samples of hippocampus (HIP), entorhinal cortex (EC), media temporal gyrus (MTG) and primary visual cortex (VCX) which have close relationship of human learning and memory. The reconstructed AD regulatory modules demonstrated that integration of the significant genes from more brain areas can enrich the information of genes and their pathways that play a prominent role in AD and improve the validity of the gene regulatory network.