An Embedded-based Feature Selection Deep Learning Architecture for Alzheimer’s Disease Detection

Aging is associated with many health problems, including Alzheimer’s Disease (AD). Currently, there is no cure for AD, and treatment decreases progression to some extent, Early AD detection may enhance the efficiency of existing therapies. A classification model using deep learning (DL) was created to establish an association between AD and DNA methylation based on the dataset collected from the GEO Omnibus database. The model was processed, quality controlled, normalized, and analyzed downstream. The most successful feature selection technique using embedded filter functions was chosen to be the final classification model after four strategies were tried. Testing compared the accuracy of the generalized improved Deep Recurrent Neural Networks model with other classification models such as CNN, RNN, and DRNN. As per the research findings, the Improved DRNN approach achieves more accurate classification and performs better than the other three models.