In the twenty-first century of the healthcare system, Alzheimer-s Disease (AD) remains the main challenge for prevention and management. In many developed countries, Alzheimer-s Disease (AD) is one of the leading causes of death, and the most progressive neurological disorder impairs memory and cognitive awareness. Researchers have a major concentration on the early diagnosis of AD in recent times due to the unavailability of its well-known causes and no permanent remedial solution for AD-infected patients.
  Artificial Intelligence (AI) technology imparts a broad range of methods to analyze huge and complex data to improve knowledge in the AD research field. There are so many machine learning algorithms explored with a special focus by the researchers to investigate varied forms of AD from the Magnetic Resonance Images (MRI) and Electroencephalogram (EEG). More specifically, AD detection using EEG achieved remarkable attention to complement traditional diagnosing methods in clinical fields, as it differentiates Healthy Control (HC) and Mild Cognitive Impairment (MCI). The below section offers different techniques, recent findings, and future challenges for machine learning in AD diagnosis research.

Key Techniques of Machine learning in Alzheimer-s Disease Detection: Some of the important machine learning techniques are highlighted here as suggestions for the effective development of AD diagnosis.
Support Vector Machine (SVM):
  •  SVM-based analysis for Alzheimer-s disease detection utilizes imaging modalities including MRI, structural MRI (sMRI), resting-state functional MRI (rs-fMRI), functional MRI (fMRI), and Diffusion Tensor Imaging (DTI).
  •  Feature selection and extraction are used by SVM to categorize control normal state of AD oMultiple kernel functions of SVM are employed for the classification of AD.
  •  Contiguous SVM, twin support vector machine, temporally structured SVM, and random forest robust SVM are some of the variants of SVM developed for the classification of Alzheimer-s disease.
  •  Ensemble-based SVM is applied for Alzheimer disease detection to improve the prediction accuracy.
Artificial Neural Network (ANN):
  •  For analyzing highly nonlinear data patterns, ANNs of machine learning are applied in Alzheimer-s disease classification.
  •  Its different forms include Random Forest (RF), Best-First decision tree (BF tree), decision tree, bagging, Multilayer Perceptron (MLP), Radial Basis Function Neural Network (RBFNN), Probabilistic Neural Network (PNN), and many more.
  •  Other approaches like Multi-Task Learning (MTL), Transfer Learning (TL), and Multi-Kernel Learning (MKL) are also utilized.
Deep Learning (DL):
  •  Deep learning, regarded as a state-of-the-art machine learning technique, has shown significant attention in the early detection and automated classification of AD.
  •  A considerable advantage of deep learning in AD is its quick progress in neuroimaging and huge-scale multimodal neuroimaging data.
  •  Deep Neural Network (DNN), Restricted Boltzmann Machine (RBM), Deep Boltzmann Machine (DBM), Deep Belief Network (DBN), Auto-Encoder (AE), Sparse AE, and Stacked AE are popular deep learning models used for Alzheimer-s disease diagnostic classification.
  •  The use of deep learning in AD conducts faster analysis with consistent accuracy than human experts.
Ensemble Methods:
  •  Ensemble methods for AD classification impart more robust, stable, and improved classification performance.
  •  Ensemble of classifiers incorporates machine learning, ANN, and deep learning models to build effective AD detection.

Current Challenges in Machine Learning for Alzheimer-s Disease Detection: Even though the noteworthy outcomes of AD diagnosis using machine learning, still it retains some unresolved challenges that need to be developed, such as,
  •  Transparency and reproducibility of deep learning models need to be addressed.
  •  Lack of efficient machine learning models for early diagnosis of Alzheimer-s disease.
  •  Discovering the optimal amalgamation of various biomarkers is essential.
  •  Partial datasets in multi-modality research for AD diagnosis must be contented.
  •  Data synthesis for AD classification needs to be more studied.

Future Findings of Alzheimer-s Disease Detection via Machine Learning: Some of the future directions of Alzheimer-s disease diagnosis using machine learning are described below to provide potential support in developing effective prevention and personalized treatment of Alzheimer-s.
  •  As a future scope, more localized pattern recognition will be enabled to identify the alterations in the brain by splitting the images into various equal parts for feature extraction and AD classification.
  •  For further enhancement in AD detection, combined with multiple modalities, instigating reinforcement learning will be developed.
  •  Using hybrid models instead of only a deep learning network will enrich the performance of Alzheimer-s disease diagnosis classification in the future.