Research Article

Stress Classification Using Brain Signals Based on LSTM Network

Table 7

Comparison of stress classification accuracies.

ReferenceNumber of subjectsNumber of electrodesLevels of stressStimulusClassification methodAccuracy (%)

[31]18322Emotional video clipsMean asymmetry scores
[37]9142 and 4SCWTSVM85.17 (2 classes)
67.06 (4 classes)
[61] 1233SCWTSVM72.3
[35]7143Multitasking, SCWTSVM77.53
Arithmetic calculations and memory
[62]2842 and 3Public speakingMLP92.85 (2 classes)
64.28 (3 classes)
[63]1012SCWTSVM97.6
[21]9142High and low altitudeFC-DNN86.62
Construction site
Proposed3542Emotional video clipsLSTM 2187.22
Proposed3542Emotional video clipsLSTM 2289.28
Proposed3542Emotional video clipsLSTM 2390.33
Proposed3542Emotional video clipsLSTM 2493.17

1Result for 50–50 training-testing data, 2result for 60–40 training-testing data, 3result for 70–30 training-testing data, and 4result for 10-fold cross validation.