A Depression Detection Model using Deep Learning and Textual Entailment
| Main Authors: | Manar Elshazly, Mohamed Hassan Haggag, Soha Ahmed Ehssan |
|---|---|
| Format: | Article Journal |
| Terbitan: |
, 2022
|
| Subjects: | |
| Online Access: |
https://zenodo.org/record/5852685 |
Daftar Isi:
- Abstract—Depression detection nowadays is essential to help in supporting depressed people. Detecting emotional disturbance is currently remarkable in people who suffer from depression, and yet for doctors and psychologists to help them in detection. Nowadays, social networks can be utilized to determine depressive content and thus depressed people. To accomplish this, twitter is used to collect the most recent tweets that is related to depression. This is done by PHQ-9 technique that classifies depression into 9 degrees. Each degree is represented by set of words. Using this classification, the model can alert users that need a have a visit to a psychiatrist or ask a psychologist as soon as possible based on their social content. The collected dataset is then trained using deep learning and then experimented with different tweets from the collected dataset to validate the model. In addition, with textual entailment, the model can determine whether the tweet is entailed or not from tweets used in the training phase, and thus will follow the same class. By combining deep learning with textual entailment, our model resulted in an improved and quicker depression detection. Index Terms—deep learning ; depression detection ; Textual entailment ; PHQ-9 ; Social Networks