Deep Learning-Based Detection of Depression and Suicidal Tendencies in Social Media Data with Feature Selection
Social media has become an essential platform for understanding human behavior, particularly in relation to mental health conditions such as depression and suicidal tendencies. Given the increasing reliance on digital communication, the ability to automatically detect individuals at risk through the...
Saved in:
| Main Authors: | , , |
|---|---|
| Format: | Article |
| Language: | English |
| Published: |
MDPI AG
2025-03-01
|
| Series: | Behavioral Sciences |
| Subjects: | |
| Online Access: | https://www.mdpi.com/2076-328X/15/3/352 |
| Tags: |
Add Tag
No Tags, Be the first to tag this record!
|
| Summary: | Social media has become an essential platform for understanding human behavior, particularly in relation to mental health conditions such as depression and suicidal tendencies. Given the increasing reliance on digital communication, the ability to automatically detect individuals at risk through their social media activity holds significant potential for early intervention and mental health support. This study proposes a machine learning-based framework that integrates pre-trained language models and advanced feature selection techniques to improve the detection of depression and suicidal tendencies from social media data. We utilize six diverse datasets, collected from platforms such as Twitter and Reddit, ensuring a broad evaluation of model robustness. The proposed methodology incorporates Cumulative Weight-based Iterative Neighborhood Component Analysis (CWINCA) for feature selection and Support Vector Machines (SVMs) for classification. The results indicate that the model achieves high accuracy across multiple datasets, ranging from 80.74% to 99.96%, demonstrating its effectiveness in identifying risk factors associated with mental health issues. These findings highlight the potential of social media-based automated detection methods as complementary tools for mental health professionals. Future work will focus on real-time detection capabilities and multilingual adaptation to enhance the practical applicability of the proposed approach. |
|---|---|
| ISSN: | 2076-328X |