Microarray-Based Classification Model of Parkinson Identification by using Firefly Algorithm-Support Vector Machine

Parkinson's disease is a disease that causes patients unable to use their motor limbs properly and correctly. More severe symptoms can be prevented by performing an early detection using tools that is able to detect Parkinson's which lead the patient to perform medical therapy to cure Parkinson's disease. This study aims to implement firefly algorithm-support vector machine to classify this blood expression either it was classified as Parkinson or normal blood expression. Firefly Algorithm was used to perform feature selection while support vector machine was used as a model classifier. We utilized three kernels for our model classification, i.e., linear kernel, polynomial kernel, and radial basis function kernel. The method that we conducted to detect Parkinson's disease was by observe the expression of blood in the human body used microarray technology from 105 individuals. According to the result, we found that there was an increase in performance after the parameter had been tuned on the linear model and the radial basis function of support vector machine model. The best model is the RBF kernel where the value of accuracy and F1-score are 0.66 and 0.67, respectively. The best model also produced value 0.66 of AUC score. It means this method can be an alternative tool to detect Parkinson's disease by transcriptome-wide scan blood sample.