Evaluation of Machine Learning Algorithms for Alzheimer's Disease Prediction Using Extensive Health Information

Early diagnosis and accurate prediction of Alzheimer's Disease are crucial to provide early intervention to improve patient outcomes. This study investigates the application of ten different supervised machine learning (ML) algorithms-LR, k-NN, DT, Gaussian Naïve Bayes, SVM, RF, Gradient Boosting, AdaBoost, and Multilayer Perceptron (MLP)-to predict Alzheimer's disease using a clinical dataset. We preprocessed our data, consisting of 2,149 samples, to remove missing values, scale feature sets, generate them into different forms with one-hot encoding, and extract features. Models were trained on 80% of the data and evaluated on a 20% test subset using accuracy, precision, recall, F1-score, and ROC-AUC. The overall results show that ensemble methods, particularly Gradient Boosting (accuracy = 95.6%, ROC-AUC = 0.986) and Random Forest (accuracy = 94.2%, ROC-AUC = 0.979), obtain better performance than other methods, demonstrating their robustness to heterogeneous biomedical data. It highlights how ML-based methodology can be applied to the early prediction of AD and the utility, especially in improving clinical decision-making and increasing diagnostic accuracy, of the ensemble models.