Integrated Deep Learning Based Optimization Algorithms for Medical Data: Fundamentals, Challenges, and Future Perspectives

Abstract

Fast and accurate disease identification is crucial for saving patients’ lives. Advances in Artificial Intelligence (AI) and Deep Learning (DL) have significantly transformed medical data analysis, enabling more precise and timely disease detection. However, the growing dimensionality and complexity of clinical data present persistent challenges, including high computational costs, memory limitations, and reduced model generalizability. Traditional DL models often underperform when processing noisy or irrelevant features, hindering their effectiveness in real-world healthcare applications. To address these limitations, researchers have begun integrating Metaheuristic Algorithms (MHs) into DL systems, optimizing performance through efficient Feature Selection (FS) and Hyperparameter Optimization (HPO). This review offers a comprehensive analysis of AI applications in healthcare, focusing on how MHs enhance deep learning models. It explores various MH types, their implementations in healthcare, and their impact on model efficiency. Additionally, the review critically evaluates existing research, highlighting challenges and limitations in integrating MHs into AI-driven healthcare solutions. Unlike prior reviews, this work emphasizes MH integration for feature selection and hyperparameter optimization in clinical DL models. It further distinguishes itself by systematically comparing recent studies across diverse medical datasets to assess real-world applicability and performance.