Deep Neural Network Classifier for Alzheimer’s Disease

Alzheimer's disease (AD) is a neurodegenerative characterized by dementia and, eventually, loss of cognitive abilities. Two histopathological features are associated with AD, neurofibrillary tangles, and amyloid-beta plaque. Both contribute to neuron cell death, dysfunction, AD pathogenesis. Current methods diagnose remain reliant on symptomatic diagnosis interviews that can be time-consuming, costly, inaccurate. Alternative such as brain imaging expensive require extensive laboratory setup for accurate results. Thus molecular-level quantitative approaches necessary. Omics datasets machine learning technology advancements have opened new avenues AD. This paper proposes using statistical principal component analysis, t-distributed stochastic neighbor embedding, Kolmogorov-Smirnov test combined Benjamini-Hochberg correction through feature selection dimensionality reduction isolate significant Furthermore, we developed models based logistic regression, random forest classifier, deep neural network (DNN) classifier predict diagnosis. Eight unique genes (TGM2, NKIRAS1, SYK, GABARAPL2, ABCC12, NDEL1, TEP1) were identified biomarkers confirmed previous works identifying prognoses' roles in After hyperparameter tuning, the DNN model showed best prediction performance among three algorithms. The preprocessed dataset demonstrated 5-fold cross-validation accuracy 0.823 AUC-ROC 0.940. Its code publicly available at https://www.kaggle.com/neobrando/ml-dnn.