Leveraging Artificial Intelligence in the Diagnosis of Neurodegenerative Disorders (Alzheimer's and Dementia) and in Managing Medication Schedules

Shawish, Amr; Mohamed, Shahed; Essam, Aya; Ahmed, Sara; Ahmed, Shimaa; Mohamed, Riham; Maher, Mahmoud; Kamal, Nourhan; Zareef, Kerols; Zidan, Youssef; Yousry, Nahla; El-Batrawy, Hoda

doi:10.21608/djis.2025.389085.1008

Leveraging Artificial Intelligence in the Diagnosis of Neurodegenerative Disorders (Alzheimer's and Dementia) and in Managing Medication Schedules"

Document Type : Original Article

Authors

¹ computer and information system, Computer science department, Damanhur university

² It Department, Faculty of computer and information system

³ computer and information system, Damanhur university

⁴ Computer and information system, Damanhur university

⁵ Assistant Lecture, Faculty of Computer and Information Science, Information Technology dep, Damanhur University.

10.21608/djis.2025.389085.1008

Abstract

The Alzheimer's disease (AD) is recognized as one of the most common neurodegenerative causes of dementia worldwide, having been characterized by progressive cognitive decline for the most part. Early diagnosis is crucial for effective management and intervention. It has been widely recognized recently that deep learning (DL) has been enlightened as a powerful tool to automate the analysis of medical images, especially when it comes to magnetic resonance imaging (MRI), to detect and classify the stages of AD. Consequently, AI-Based Alzheimer’s Diagnostic and Assistance System (AIDAS) leverages artificial intelligence (AI) and deep learning technologies to facilitate the automated identification, classification, and monitoring of AD progression through sophisticated analysis of neuroimaging data. Furthermore, AIDAS integrates intelligent assistance features such as personalized medication reminders and cognitive support tools, thereby enhancing patient adherence, optimizing clinical decision-making, and contributing to a more proactive and precision-driven approach to neurodegenerative disease management. The investigation employed a dataset consisting of 80,000 MRI brain scans acquired along the brain's z-axis, with class labels derived from the Clinical Dementia Rating (CDR) guidelines. This dataset includes 461 people and was derived from the Open Access Series of Imaging Study (OASIS). The experimental results underscore that the proposed ordinal EfficientNetB0 architecture outperforms other models in fundamental structural tasks, characterized by minimal computational overhead, reduced susceptibility to overfitting, efficient memory utilization, and effective temporal regulation. Furthermore, the developed Convolutional Neural Network (CNN) model demonstrated exceptional performance, achieving classification accuracies of 99.5% for AD stage identification.

Keywords