Scholarlink

<p>Alzheimer’s Disease (AD) is an incurable neurodegenerative disorder that affects memory, behaviour, and cognitive abilities. Early diagnosis is crucial for managing disease progression and improving deteriorating quality of life. The objective of this systematic review is to evaluate the role of Artificial Intelligence (AI) in AD diagnosis and to analyze the existing AI techniques, datasets, and evaluation metrics. The article references sources from various platforms, including PubMed, ScienceDirect, Web of Science, ACM Digital Library, and IEEE Xplore. After a thorough evaluation of various models, existing research gaps are identified. The current research utilizes datasets such as Alzheimer’s Disease Neuroimaging Initiative, Open Access Series of Imaging Studies, and Kaggle, and key modalities include neuroimaging, biomarkers, speech analysis, transcriptomics, clinical records, and genotype. Although the diagnostic accuracy of models is high, several challenges remain, including the use of small, imbalanced datasets and limited adoption of vertical federated learning and federated transfer learning. The identified research gaps include client heterogeneity, limited computational power, limited dataset size, lack of interpretability, and inconsistencies in evaluation metrics. Next, the future research directions are outlined including the use of more multimodal data and the development of frameworks with enhanced data-sharing mechanisms and exploration of the underutilized FL paradigm to improve generalizability and clinical implementation in AD diagnostic systems. Unlike existing review articles, this systematic review provides comprehensive coverage of hyperparameters, data distribution, and Optimizer analysis.</p>

错误：搜索内容不能为空，请输入英文关键词
错误：关键词超出字数限制，请精简
高级检索

Computational Methods for Alzheimer’s Disease Diagnosis: A Systematic Review of Deep Learning and Federated Learning with Clinical and Privacy Perspectives

  • 期刊论文

  • 发表日期：

    2026年6月29日

查看全文
  • Yashi Tyagi,
  • Robin Singh,
  • Seema Gulati,
  • Sandeep Kumar

摘要

Alzheimer’s Disease (AD) is an incurable neurodegenerative disorder that affects memory, behaviour, and cognitive abilities. Early diagnosis is crucial for managing disease progression and improving deteriorating quality of life. The objective of this systematic review is to evaluate the role of Artificial Intelligence (AI) in AD diagnosis and to analyze the existing AI techniques, datasets, and evaluation metrics. The article references sources from various platforms, including PubMed, ScienceDirect, Web of Science, ACM Digital Library, and IEEE Xplore. After a thorough evaluation of various models, existing research gaps are identified. The current research utilizes datasets such as Alzheimer’s Disease Neuroimaging Initiative, Open Access Series of Imaging Studies, and Kaggle, and key modalities include neuroimaging, biomarkers, speech analysis, transcriptomics, clinical records, and genotype. Although the diagnostic accuracy of models is high, several challenges remain, including the use of small, imbalanced datasets and limited adoption of vertical federated learning and federated transfer learning. The identified research gaps include client heterogeneity, limited computational power, limited dataset size, lack of interpretability, and inconsistencies in evaluation metrics. Next, the future research directions are outlined including the use of more multimodal data and the development of frameworks with enhanced data-sharing mechanisms and exploration of the underutilized FL paradigm to improve generalizability and clinical implementation in AD diagnostic systems. Unlike existing review articles, this systematic review provides comprehensive coverage of hyperparameters, data distribution, and Optimizer analysis.