A novel fast detection algorithm for depression based on 3-channel EEG signals

Medically unexplained symptoms (MUS) are an emerging field in current research. Among middle-aged and elderly patients, most MUS symptoms are mainly caused by depression, but early symptoms do not meet the international somatization standards, which delays treatment. Therefore, developing a rapid auxiliary diagnosis method is of great significance. This paper proposes a novel model for identifying depression based on 3-channel electroencephalogram (EEG) signals from the prefrontal lobe of the human brain. For the collected resting-state EEG signals, variational mode decomposition (VMD) is first used for signal decomposition, and the power spectrum is employed to select intrinsic mode function (IMF) components. After extracting energy features via sample entropy, LightGBM is adopted for classification, with a classification accuracy of 97.42%. Through comparative experiments, the model proposed in this paper achieves a balance between high accuracy and timeliness. This is conducive to the development of a depression detection system based on portable real-time electroencephalography (EEG), and provides a solution for EEG signal devices in real-time depression detection and pre-triage of patients with Medically Unexplained Symptoms (MUS).