EEG-based Mild Depressive Detection using Feature Selection Methods and Classifiers

• The combination of feature selection method Greedy-Stepwise (GSW) based on Correlation Features Selection (CFS) and classification algorithm KNN can achieve the optimal performance for mild depression detection.• Fewer EEG channels: FP1, FP2, F3, O2, T3 with linear features may be a good choic e for portable device and auxiliary diagnosis of mild depression.• Classification accuracy above 91% and AUC above 0.950, these results are better than some existing studies.

Source: Computer Methods and Programs in Biomedicine - August 17, 2016 Category: Bioinformatics Authors: Xiaowei Li, Bin Hu, Shuting Sun, Hanshu Cai Source Type: research