Research Square (Research Square), Journal Year: 2025, Volume and Issue: unknown

Published: March 27, 2025

Drug classification and target identification are crucial yet challenging steps in drug discovery. Existing methods often suffer from inefficiencies, overfitting, limited scalability. Traditional approaches like support vector machines XGBoost struggle to handle large, complex pharmaceutical datasets effectively. Deep learning models, while powerful, face challenges with interpretability, computational complexity, generalization unseen data. This study addresses these limitations by introducing a novel framework: optSAE+HSAPSO. framework integrates stacked autoencoder (SAE) for robust feature extraction hierarchically self-adaptive particle swarm optimization (HSAPSO) algorithm adaptive parameter optimization. combination delivers superior performance across various metrics. Experimental evaluations on DrugBank Swiss-Prot demonstrate that optSAE+HSAPSO achieves high accuracy of 95.52%. Notably, it exhibits significantly reduced complexity (0.010 seconds per sample) exceptional stability (±0.003). Compared state-of-the-art methods, the offers higher accuracy, faster convergence, greater resilience variability. Furthermore, ROC convergence analyses confirm its robustness capability, maintaining consistent both validation datasets. By leveraging advanced techniques, efficiently handles large sets diverse data, making scalable adaptable solution real-world discovery applications. However, method's is dependent quality training fine-tuning may be necessary high-dimensional Despite limitations, demonstrates transformative potential, reducing overhead improving reliability. work advances field informatics presenting reliable efficient identification. These findings open promising avenues future research, including extending other domains such as disease diagnostics or genetic data classification, ultimately accelerating development process.

Language: Английский