Machine Learning Model for Computer-Aided Depression Screening among Young Adults Using Wireless EEG Headset

Abstract

Depression is a disorder that if not treated can hamper the quality of life. EEG has shown great promise in detecting depressed individuals from depression control individuals. It overcomes the limitations of traditional questionnaire-based methods. In this study, a machine learning-based method for detecting depression among young adults using EEG data recorded by the wireless headset is proposed. For this reason, EEG data has been recorded using an Emotiv Epoc+ headset. A total of 32 young adults participated and the PHQ9 screening tool was used to identify depressed participants. Features such as skewness, kurtosis, variance, Hjorth parameters, Shannon entropy, and Log energy entropy from 1 to 5 sec data filtered at different band frequencies were applied to KNN and SVM classifiers with different kernels. At AB band (8-30 Hz) frequency, 98.43 ± 0.15% accuracy was achieved by extracting Hjorth parameters, Shannon entropy, and Log energy entropy from 5 sec samples with a 5-fold CV using a KNN classifier. And with the same features and classifier overall accuracy = 98.10 ± 0.11, NPV = 0.977, precision = 0.984, sensitivity = 0.984, specificity = 0.976, and F1 score = 0.984 was achieved after splitting the data to 70/30 ratio for training and testing with 5-fold CV. From the findings, it can be concluded that EEG data from an Emotiv headset can be used to detect depression with the proposed method.

Figure 1

Data acquisition using Emotiv Epoc+ and PHQ9 screening tool.

Figure 2

Timeline for each recording session.

Figure 3

Channel locations, Emotiv headset, and channels at different regions of the brain.

Figure 4

Raw data from control and depressed group.

Figure 5

Brain maps of control and depressed participants at different frequency points.

Figure 6

Segmentation and signal processing.

Figure 7

Filtered data at different frequencies.

Figure 8

Creating feature matrix by extracting features from filtered data.

Figure 9

Classifying depressed and depression control participants using SVM and KNN.

Figure 10

Data splitting and cross-validation.

Figure 11

Confusion matrix.

Figure 12

Classification accuracies at different sample lengths using skewness feature.

Figure 13

Classification accuracies at different sample lengths using kurtosis feature.

Figure 14

Classification accuracies at different sample lengths using Hjorth parameters.

Figure 15

Classification accuracies at different sample lengths using variance feature.

Figure 16

Classification accuracies at different sample lengths using Shannon entropy and log energy entropy features.

Figure 17

Classification accuracies at different sample lengths using Hjorth parameters, Shannon entropy, and Log energy entropy features.

Figure 18

Classification accuracies at different sample lengths using Shannon entropy, Log energy entropy, and variance features.

Figure 19

Classification accuracies at different sample lengths using skewness and kurtosis features.

Figure 20

Classification accuracies at different sample lengths using all features.

Figure 21

Classification accuracy using different sub-bands.

Figure 22

Classification accuracy using different combinations of sub-bands.

Figure 23

Training, testing, and overall accuracy of different combination of bands using fine KNN classifier.

Figure 24

Training, testing, and overall accuracy of different brain regions using fine KNN classifier.