Deep Learning Models in Early Diagnosis of Alzheimer’s Disease: A Systematic Review of Current Applications and Challenges

Improvements in health metrics and expected longevity, particularly in developed countries, have resulted in population growth and a rise in age-related illnesses, such as Alzheimer’s disease (AD). Consequently, early diagnosis of Alzheimer’s disease is crucial for halting its progression at the early stages. Considering the increasing application of artificial intelligence techniques in intelligent healthcare, this paper provides a comprehensive review of the current utilization of deep learning models on neuroimaging data for the early diagnosis of Alzheimer’s disease, while also comparing their features, advantages, and limitations. The paper focuses on the analysis of deep learning models, including Convolutional neural networks (CNNs), Recurrent neural networks (RNNs), Deep neural networks (DNNs), and Deep polynomial networks (DPNs). In addition to the technique review, this paper discusses the challenges and proposes future research directions for the early diagnosis of Alzheimer’s disease, such as insufficient dataset size, interpretability, transparency, and integrating different biomarkers.