Detection of brain neurological disorders with Electroencephalogram signals using Random Forest classifier comparing with Linear Discriminant Analysis classifier

Abstract

The purpose of this research is to determine whether or not the Random Forest (RF) algorithm is superior to the Linear Discriminant Analysis (LDA) method in terms of accuracy, recall, precision, and specificity when it comes to diagnosing brain and neurological disorders with electroencephalogram (EEG) signals (LDA). EEG recordings, which reflect the brain activity of 20 samples from each group, are analysed with RF and LDA to determine whether or not the samples have epileptic activity. The selection of prominent features from EEGs through the application of the notion of mutual information is the innovative aspect of this study. The RF method produced results with an accuracy of 95.85%, recall of 91.33%, precision of 88.28%, and specificity of 96.99%, while the LDA approach produced results with an accuracy of 68.84%, recall of 42.21%, precision of 31.43%, and specificity of 76.33%. Since p0.05, we may say that there is a substantial difference between the two groups. When it comes to accurately diagnosing neurological illnesses in the brain, RF classifiers are known to be much more accurate than LDA classifiers.