Mental workload recognition using ECG and machine learning in simulated flight tasks

Effective mental workload recognition is of great significance for improving task performance and reducing accidents. Although prior research has achieved approximately 95% accuracy using electroencephalography (EEG), it is difficult to transplant into actual task scenarios due to the low portability of the device. Here, we introduce a mental workload recognition solution to give consideration to high recognition accuracy and portability. Heart rate variability (HRV) was extracted from the electrocardiogram (ECG) signals of 26 participants during simulated flight tasks, and the sensitive features were screened out using the generalized linear mixed model. Then, the three mental workload levels were classified and evaluated in combination with the machine learning method. Our solution achieved an accuracy of 98% for subject-independent mental workload recognition.