Improving Early Detection of Alzheimer's and Parkinson's Diseases Through Novel Ant Colony Optimization-Based Feature Selection

Neurological disorders such as Alzheimer's Disease (AD) and Parkinson's Disease (PD) pose significant global health challenges, affecting millions worldwide with projected doubling of cases by 2040. Early and accurate prediction of these disorders is crucial for timely interventions that can slow disease progression and improve patient outcomes. This study introduces a novel Soldier-Enhanced Ant Colony Optimization (SACO) algorithm for feature selection, tailored for high-dimensional medical feature selection. SACO utilizes two specialized ant populations: worker ants that exploit established pheromone trails for steady feature selection, and soldier ants that introduce controlled exploration to discover novel feature combinations. Using publicly available datasets from Kaggle containing 2,149 AD instances and 2,105 PD instances, SACO was benchmarked against Recursive Feature Elimination (RFE), Backward Feature Selection (BFS), Whale Optimization (WO), and Artificial Bee Colony (ABC) algorithms. Feature subsets were evaluated using Random Forest, XGBoost, Logistic Regression, K-Nearest Neighbors, and Stacked Ensemble classifiers. SACO-selected features achieved good performance, with Random Forest and XGBoost attaining 94.96% and 94.84% accuracies (AUC=0.975 and 0.95) for AD and 92% and 92% accuracies (AUC=0.971 and 0.964) for PD respectively. Statistical validation using McNemar's test and Friedman analysis confirmed significant performance differences (p<0.001). SACO's inclusion of comprehensive cardiovascular and metabolic markers alongside cognitive assessments provided a holistic predictive signature, outperforming conventional methods by 8-12% in accuracy. These findings demonstrate SACO's potential for advancing personalized medicine through earlier, more reliable neurological disorder predictions.