JMIR Publications

ABSTRACT

Background:

As a common mental disease, depression seriously affects people's physical and mental health. According to the statistics of the World Health Organization (WHO), depression is a major reason of suicide and self-harm events in the world. Therefore, strengthening the depression detection can effectively reduce the occurrence of the suicide or self-harm events, so as to save more people and families. With the development of computer technology, some researchers are trying to apply Natural Language Processing (NLP) techniques to detect depressed people automatically. Many existing feature engineering methods for depression detection are based on emotional characteristics, but these methods do not consider the high-level emotional semantic information. The current deep learning methods for depression detection cannot accurately extract the effective emotional semantic information.

Objective:

In this paper, we propose an Emotion based Attention Network (EAN), including a semantic understanding network (SUN) and an emotion understanding network (EUN), which can capture the high-level emotional semantic information effectively to improve the depression detection task.

Methods:

The SUN module is used to capture the contextual semantic information. The EUN module is used to capture the emotional semantic information. There are two units in the EUN module, including a positive emotion understanding unit and a negative emotion understanding unit, which are used to capture the positive emotional information and the negative emotional information, respectively. We further propose a dynamic fusion strategy in EUN module to fuse the positive emotional information and the negative emotional information.

Results:

We evaluate our method on Reddit dataset. The experimental results show that the proposed EAN model achieves accuracy, precision, recall, F-measure of 91.30%, 91.91%, 96.15%, 93.98%, which achieves comparable results compared with state-of-the-art methods.

Conclusions:

The experimental results show that our model is competitive with the state-of-the-art models. The SUN module, the EUN module and the dynamic fusion strategy are effective modules for depression detection. And the experimental results verify that the emotional semantic information is effective in depression detection.