Muften, Haider Ali (2026) Predicting Alzheimer's Disease Using Artificial Intelligence Techniques. CENTRAL ASIAN JOURNAL OF THEORETICAL AND APPLIED SCIENCE, 7 (3). pp. 158-169. ISSN 2660-5317

Text CAJOTAS_Predicting+Alzheimer's+disease.pdf - Published Version Download (717kB)

Abstract

Alzheimer disease (AD) is a progressive neurodegenerative disorder beginning with the accumulation of pathological proteins in brains and ultimately leading to neuronal death. Alzheimer's disease is among the most severe of cases that have a significant deterioration in cognitive ability with particular emphasis on detrimental effects to memory, intellect and more general behavioral functions. How: There is no cure at the moment, but researchers are working tirelessly in hopes of finding one. The immediate need for early stage diagnosis and manifestation of biomarkers has streamlined the therapeutic algorithms in terms of potential drug trials & preventive medication regimens, instituted at a very early developmental phase. Electroencephalography (EEG) is simple, faster and cost-effective non-invasive technique which can be used as adjunct for automation of Alzheimer's disease diagnosis. It merits inption that epoch length of segment EEG signals data might impact the performance for classification. To tackle this issue, we presented a device-free diagnostic EEG framework, where the ideal segment length estimation for classification is obtained using machine learning and deep learning-based approaches. It consists of the data collection of EEG, preprocessing by removing noise, and segmentation in time axis. In this work, we run a comparison of using deep learning models (multilayer neural networks and convolutional neural networks) to the more traditional machine learning models. Model: Training (logospheric regression, decision tree, random forest, gradient enhancement, AdaBoost, XGBoost; CNN and MLP); Classification; Evaluation The accuracy we obtained using open data Kaggle set is 0.83%, 96,7%. and 99,3% respectively. We tested the proposed models on an entirely novel application of identifying frontotemporal dementia and achieved substantial advances relative to previous publications. Moreover, we performed several analyses and graphically presented extracted categories contents to justify the developed model. The study will set a standard in the realm of neurological disorder research and one that will support future researchers and technical experts focused on this field.

Item Type:	Article
Subjects:	A General Works > AI Indexes (General)
Depositing User:	admin eprints
Date Deposited:	29 Jun 2026 07:34
Last Modified:	29 Jun 2026 07:34
URI:	http://eprints.umsida.ac.id/id/eprint/16677

Actions (login required)

View Item