Performance analysis of machine learning models on Alzheimer's prediction and meta-heuristic algorithms in feature selection

The prediction of Alzheimer’s disease remains a significant challenge in data science and healthcare. This research explores the application of metaheuristic algorithms, including the Genetic Algorithm, Grey Wolf Optimizer, and Whale Optimization Algorithm, for optimal feature selection. These methods aimed to improve model accuracy, reduce computational complexity, and identify the most critical features in the dataset. Initially, various machine learning models, such as Random Forest, XGBoost, LightGBM, CatBoost, and Logistic Regression, were employed to evaluate the dataset. Metaheuristic algorithms were then applied for feature selection, extracting the most relevant features. Results revealed that combining these algorithms with machine learning models significantly improved predictive performance. Specifically, the LightGBM model, paired with features selected by the Grey Wolf Optimizer, and the Bagging model, combined with the Whale Optimization Algorithm, achieved the highest accuracy in predicting Alzheimer’s disease. This study highlights the importance of feature selection in enhancing the performance of machine learning models for complex, high-dimensional datasets. It provides a foundation for future research in leveraging optimization techniques to improve medical predictions and address other cognitive challenges effectively.

کلیدواژه ها:

Alzheimer’s disease ، machine learning ، feature selection ، dimensionality reduction ، metaheuristic algorithms