Optimizing Gradient Boosting Machine for Alzheimer’s Disease Prediction using Cormorant Swarm Intelligence (CSO-GBM)

Background: Alzheimer’s Disease (AD) is a background of progressive neurodegenerative disorder which affects cognitive function and memory. In the traditional diagnostic approaches, there exists a lot of technical issues such as, hyperparameter tuning challenges, data imbalance and limited generality of predictive models result into the late detection, complex data interpretation and low classification performance. Objectives: The objectives are to develop an improved classification framework by enhancing the Gradient Boosting Machine (GBM) with a biologically inspired optimization algorithm and apply it to accurately and early predict AD. The desired result to accomplish this is to have a robust and adaptive learning process that can handle the above issues, including sub optimal hyperparameter selection, overfitting and sensitivity to data imbalance.Methods: The proposed model is Cormorant Swarm Optimized Gradient Boosting Machine (CSO-GBM) which is inspired from the cormorant’s cooperative hunting and adaptive foraging behavior. The hyperparameter refinement is done efficiently by balancing the global exploration and local exploitation using the optimization algorithm. In the proposed work evaluation on the OASIS Alzheimer’s MRI dataset and compared with existing models, INN-MT and FDCNN-AS. Findings: Results showed that the CSO-GBM model provided a high classification accuracy result of 82.39% and the highest F-measure of 83.65% as compared to benchmark models. It had shown better performance in reducing false positives and false negatives; at the same time, it accommodates high adaptability, robustness, and computational efficiency over complex data distribution. Novelty: This is the first research to explore swarm intelligence derived from the cormorant for optimizing GBM for predicting Alzheimer’s Disease. Dynamic convergence, high generalization ability and the possibility of integration into medical diagnostic frameworks were ensured by an adaptive search mechanism; thus, the algorithm provides an excellent basis for a computationally efficient and clinically reliable classification model. Keywords: Alzheimer’s Disease Classification, Bio-Inspired Optimization, Cormorant Optimization, GBM, Feature Selection, Healthcare