The number of medical applications with large datasets that require
great speed and accuracy is continually growing. A large number of features in
medical datasets is one of the most critical issues in data classification and prediction models. Furthermore, irrelevant and redundant features have also harmed the
complexity and functioning of data classification systems. Feature selection is a
reliable dimensionality reduction strategy for identifying a subset of valuable and
non-redundant features from massive datasets. This paper reviews the state-of-the-art
feature selection techniques on medical data in the last five years.