METAHEURISTIC ALGORITHMS: A PARADIGM SHIFT IN DIMENSIONALITY REDUCTION IN HEALTHCARE INFORMATICS

ABSTRACT

Healthcare informatics leverages on massive data generated at an expeditious rate from diverse domains in healthcare. Such data culminates in redundant, noisy, complex and high dimensional attributes which impose a load on the learning algorithms employed for disease diagnosis, prognosis, prediction and treatment planning. Obtaining relevant information from the chunk of data generated is a challenge in healthcare informatics, hence the need for dimensionality reduction techniques. Traditional dimensionality reduction techniques while effective, may not handle noisy and complex data in some domain-specific healthcare scenarios. As such, the choice of adopting metaheuristics as dimensionality reduction techniques in healthcare informatics becomes important due to their optimization capability in selecting optimal features in data sets and their effect on classifier performance. This paper aims to discuss an overview of some state-of-the-art metaheuristic algorithms popularly adopted as dimensionality reduction techniques in healthcare informatics. A brief review of recent research related to metaheuristic algorithms in healthcare informatics was carried out. This work concludes by presenting the identified trends, potentials, challenges, and prospects of metaheuristics in healthcare informatics.

Keywords: Metaheuristics, Dimensionality Reduction, Healthcare Informatics