Feature Selection Methods in Big Medical Databases: A Comprehensive Survey

Abstract

Medical science is rapidly evolving alongside the continuous growth of medical data recording systems, resulting in massive and diverse datasets. While these data offer valuable opportunities, their high volume and complexity create significant challenges in processing, particularly in tasks such as classification and clustering. Feature selection has emerged as a critical strategy to overcome these issues by improving efficiency, accuracy, interpretability, and scalability. This article presents a comprehensive survey of feature selection methods with a dedicated focus on medical data, providing a structured categorization into filter, wrapper, and embedded approaches. It further reviews evaluation criteria, highlights strengths and limitations of each category, and discusses their relevance through examples from real-world medical applications. By combining theoretical perspectives with practical insights, this work contributes a clear roadmap for researchers in healthcare informatics, emphasizing that effective feature selection can substantially enhance medical data analysis and support future advancements in the field.