Detection Of Depression Via Analyzing The Electroencephalograms Acquired Under Various Activities

Abstract

The total number of the patients with the depression continues to grow in the recent years. The early detection of the depression is conducive to the timely treatment of the patients. This paper mainly studies whether the people are suffered from the depression or not via analyzing the electroencephalograms acquired under various daily activities. In particular, four patients are suffered from the depression and four people are healthy. They are asked to perform seven activities with the high concentration. Here, the conducted activities are the drawing activity, the eating activity, the doing computer exercises activity, the playing electronic games activity, the reading activity, the playing with the toys activity and the watching the television activity. The electroencephalograms are collected when these activities are conducted. Then, the electroencephalograms are filtered with the passbands of the filtered electroencephalograms being between 100Hz and 150Hz. Next, the empirical mode decomposition is performed. The first four intrinsic mode functions are used to extract the features. Finally, the back propagation neural network, the support vector machine and the random forest are used to classify between the depression patients and the healthy people. It is found that the highest classification accuracy is 89.27%. Therefore, it can be concluded that the electroencephalograms acquired under various activities can be used to detect whether a person has suffered from the depression or not.