A COMPARATIVE STUDY OF MACHINE LEARNING MODELS FOR PREDICTING THE RISK OF ALZHEIMER’S DISEASE
Loading...
Files
Date
Authors
Journal Title
Journal ISSN
Volume Title
Publisher
Universiti Malaysia Sarawak (UNIMAS)
Abstract
Description
The main objective of this project is to compare machine learning models to predict the risk of Alzheimer’s disease. As the aging population continues to grow, the number of Alzheimer’s disease patients is becoming increasingly concerning. This trend is often associated with unhealthy lifestyle patterns, which may exacerbate the risk. Thus, this project uses machine learning to predict Alzheimer’s disease risk, enabling individuals to assess their risk and take proactive steps to reduce it. This project proposed a classification approach for this comparative study to predict the risks. This is implemented using the Knowledge Discovery in Databases (KDD) methodology with steps including data selection, pre-processing, transformation, data mining, and evaluation. The proposed prediction models are logistic regression, decision tree, support vector machine (SVM), CatBoost, XGBoost, random forest, and artificial neural network (ANN). In this study, the dataset collected from Kaggle underwent preprocessing involving data cleaning, feature selection, data engineering, and data standardization. It was then divided into training (80%) and testing (20%) sets using the holdout method. Next, these various models were compared using evaluation metrics derived, such as accuracy, precision, recall, F1 score, and sensitivity from the confusion matrix to identify the best-performing models. From this study, the best-performing model is the XGBoost model. It shows the highest performance among all the machine learning algorithms with accuracy at 95.3%, precision at 95.5%, recall at 94.3%, and F1-score at 94.9%.