An Early Detection of Alzheimer’s Disease Using Machine Learning Techniques

Date
2022
Research Supervisor
Editor
Journal Title
Volume
Issue
Journal Title
Journal ISSN
Volume Title
Abstract
Alzheimer’s disease is a brain disorder that increasingly reduces one’s capacity for thinking, remembering, and performing even the most fundamental tasks. It is an irreversible, progressive condition that slowly destroys memory and thinking skills and eventually the ability to carry out the simplest tasks. The symptoms of Alzheimer’s vary from person to person and can be mild in the early stages but worsen over time as the disease progresses. The majority of people with late-onset symptoms often start to exhibit symptoms in their mid-60s. Dementia has the potential to profoundly alter a person’s quality of life. If this disease can be identified early on, brain damage can be avoided. Early detection makes it easier to receive care and treatment in the future. It helps people plan ahead while they are still able to make important decisions about their care and support needs, as well as financial and legal issues. The major objective of this research is to create a machine-learning model that can identify Alzheimer’s disease at an early stage. It will determine if someone has Alzheimer’s disease or not. In this research, two different datasets were used: the OASIS dataset and the ADNI dataset. This research has generally been conducted using three different approaches. A few machine learning models, such as SVM (Support Vector Machine), Decision Tree, Logistic Regression, Random Forest, KNN, and others, have first been applied separately to the ADNI dataset and the OASIS dataset. According to the research done on each dataset, the Random Forest classifier after oversampling is the best-performing model with 96% on the ADNI dataset, and the best-performing models on the OASIS dataset are the SVM (Support Vector Machine) and Logistic Regression with 91% accuracy without any imbalance handling. When working with the ADNI dataset alone, two different approaches were used. First, the dataset was trained and tested using 5 classes: Normal (CN), Early Mild Cognitive Impairment (EMCI), Late Mild Cognitive Impairment (LMCI), Subjective Memory Complaint (SMC), and Alzheimer’s disease (AD). Then, as an alternative strategy, the ADNI dataset was divided into 3 classes (CN, MCI, and AD) and trained and tested using those. In the ADNI 3-class model, there are three best-performing models: KNN, SVM, and Decision tree with 97% accuracy after oversampling. Finally, an Explainable AI technique, Lime, is used in this study to better explain the behavior of trained models. In this study, a variety of methods have been used to determine the most precise way to predict Alzheimer’s disease using machine learning.
Description
Keywords
Citation
Department Name
Electrical and Computer Engineering
Publisher
North South University
Printed Thesis
DOI
ISSN
ISBN