Depression Recognition Based on Electrocardiogram

Depression is a major psychiatric concern and severely impacts the quality of life. Currently, questionnaires and physiological signals are the main methods to assess depression. This paper proposed to achieve depression detection based on electrocardiogram (ECG) signals from realistic scenes. Forty-one college students took part in the entire experiment. Thirty-nine indicators of the autonomic nervous system were calculated, including the time domain, frequency domain, and non-linear features. Then, five machine learning classifiers were applied to build recognition of the depressed state. The support vector machine achieved the best performance, which had an accuracy of 67.00 <sup xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink">%</sup> in validation and 70.00% accuracy in the independent-subjects test.