An automatic diagnostic network using skew-robust adversarial discriminative domain adaptation to evaluate the severity of depression

Detecting depression tendency with multimodal features

BACKGROUND AND OBJECTIVE

Depression can severely impact physical and mental health and may even harm society. Therefore, detecting the early symptoms of depression and treating them on time is critical. The widespread use of social media has led individuals with depressive tendencies to express their emotions on social platforms, share their painful experiences, and seek support and help. Therefore, the massive available amounts of social platform data provide the possibility of identifying depressive tendencies.

METHODS

This paper proposes a neural network hybrid model MTDD to achieve this goal. Analysis of the content of users' posts on social platforms has facilitated constructing a post-level method to detect depressive tendencies in individuals. Compared with existing methods, the MTDD model uses the following innovative methods: First, this model is based on social platform data, which is objective and accurate, can be obtained at a low cost, and is easy to operate. The model can avoid the influence of subjective factors in the depressive tendency detection method based on consultation with mental health experts. In other words, it can avoid the problem of undisclosed and imperfect data in depressive tendency detection. Second, the MTDD model is based on a deep neural network hybrid model, combining the advantages of CNN and BiLSTM networks and avoiding the problem of poor generalization ability in a single model for depression tendency recognition. Third, the MTDD model is based on multimodal features for learning the vector representation of depression-prone text, including text features, semantic features, and domain knowledge, making the model more robust.

RESULTS

Extensive experimental results demonstrate that our MTDD model detects users who may have a depressive tendency with a 95% F1 value and obtained SOTA results.

CONCLUSIONS

Our MTDD model can detect depressive users on social media platforms more effectively, providing the possibility for early diagnosis and timely treatment of depression. The experiment proves that our MTDD model outperforms many of the latest depressive tendency detection models.

A Convenient and Low-Cost Model of Depression Screening and Early Warning Based on Voice Data Using for Public Mental Health

Depression is a common mental health disease, which has great harm to public health. At present, the diagnosis of depression mainly depends on the interviews between doctors and patients, which is subjective, slow and expensive. Voice data are a kind of data that are easy to obtain and have the advantage of low cost. It has been proved that it can be used in the diagnosis of depression. The voice data used for modeling in this study adopted the authoritative public data set, which had passed the ethical review. The features of voice data were extracted by Python programming, and the voice features were stored in the format of CSV files. Through data processing, a big database, containing 1479 voice feature samples, was generated for modeling. Then, the decision tree screening model of depression was established by 10-fold cross validation and algorithm selection. The experiment achieved 83.4% prediction accuracy on voice data set. According to the prediction results of the model, the patients can be given early warning and intervention in time, so as to realize the health management of personal depression.