Feature selection in medical records is necessary because the data usually contains many irrelevant features and noise.Electronic Health Record, abbreviated as EHR, makes it possible to analyze large amounts of medical data. A Genetic Algorithm is widely used for feature selection because it has the ability or potential for global optimization of theselected features. Genetic Algorithm-based methods include many iterations (generations) in the crossover process,and mutation can produce new individuals because the Genetic Algorithm adopts a fitness value to represent how''good'' the resulting individual is. The problem with heuristic algorithms is those simple genetic algorithms are notfor processing high- dimensional data. Genetic algorithms in solution search techniques always get local optimumsolutions which can cause failure to obtain optimal solutions during random searches. From these limitations,developing and improving genetic algorithms for feature selection on clinical data is necessary. First, sort the featuresbased on the feature evaluation criteria to exclude irrelevant features through the fitness process in the evaluation withthe accuracy value of the Support Vector Machine calculation. This way reduces the number of features and results inoptimal features. Then to get the optimal solution, it is necessary to optimize the subset features that have been selectedusing a machine learning algorithm that determines the best parameters using a genetic algorithm.

Link publikasi : https://www.apcore-inc.org/_files/ugd/183efc_8dd695dffdb549f9bd16b7eca347c941.pdf