A Human-Machine Hybrid Approach for Diagnosing Depression Based on Convolutional Neural Networks

This paper discusses a new method that combines electroencephalography (EEG) and electrooculography (EOG) to improve the accuracy of depression diagnosis. By analyzing the differences in EEG signals between open and closed eye states, the study found significant differences in brain activity between healthy individuals and those with depression under different eye conditions. Based on these findings, a new diagnostic process is proposed, which first uses EOG signals to determine the subject's eye state, and then selects appropriate EEG features for analysis according to the different eye states. Additionally, the paper introduces the concept of human-machine hybrid intelligence. This approach combines the diagnostic results from deep learning algorithms with expert opinions from doctors, further enhancing diagnostic accuracy. The experiments demonstrate that when we consider the state of the eye and use the human-machine hybrid model, the diagnostic performance of depression is improved, compared to diagnoses that do not consider eye states and expert opinions.