Paper Title
Empowering Dementia Diagnosis: A Machine Learning-Driven Automated System

Abstract
Dementia, a neurodegenerative disease, significantly impairs cognitive abilities and is often not diagnosed until the later stages of disease progression. This delayed diagnosis results in missed early intervention, support opportunities and difficulty implementing appropriate care strategies. To address this problem, researchers have proposed automated diagnostic systems that utilize machine learning methods using electronic health records. Inspired by these advances, we have developed an automated diagnostic system comprising two components. The first component extracts valuable information from the dataset, while the second component uses this extracted information to make accurate decisions.Our approach uses linear discriminant analysis (LDA) to extract features from the dataset and classify subjects as demented or healthy, resulting in the name LDA-DT for the proposed diagnostic system. The hyperparameters of the decision tree were fine-tuned using the grid search algorithm. To validate the efficacy of our diagnostic system, we conducted two types of experiments. In the first experiment, we evaluated the performance of conventional decision trees based on LDA. In contrast,we used an optimized decision tree using the grid search algorithm in the second experiment. We accurately evaluated the performance of the developed LDA-DT model using several validation metrics. Our proposed model achieved a remarkable accuracy of 97.77%, demonstrating its potential for accurate dementia classification and its promising impact in improving the diagnostic process for this complex disease. Keywords - Feature Extraction, Decision Tree, Classification, Dementia Prediction