Unveiling the role of social media in mental health: A GAN-based deep learning framework for suicide prevention

In recent years, there has been a significant increase in user participation on social networking media sites. These platforms generate vast amounts of diverse data that have a substantial impact on the mental health of the general public. Suicide, being a leading cause of death globally, has drawn the attention of researchers. The World Health Organization estimates that around 800,000 people died by suicide in 2019, with a significant portion falling between the ages of 15 and 29. The COVID-19 pandemic has further exacerbated the issue due to social isolation. Traditionally, the study of suicide has focused on physiological aspects using questionnaires and in-person settings. However, the effectiveness of such approaches is hindered by societal stigma. To address this, we propose a method that combines Generative Adversarial Networks (GANs) with various deep learning algorithms, including Convolutional Neural Networks (CNNs), Long Short-Term Memory (LSTM) networks, Bidirectional Long Short-Term Memory (BI-LSTM) networks, and Gated Recurrent Units (GRUs). We also employ multiple feature engineering techniques, such as GloVe, Word2Vec, Fasttext, and Weighted Average Fusion (GloVe+Word2Vec+Fasttext). Our results demonstrate that the combination of CNN with Fasttext yields impressive recall, precision, and accuracy measures, with caps of 95%, 95%, and 92%, respectively. This research contributes to the field of suicide prevention by utilizing deep learning models and feature engineering methods to analyze social media data. By leveraging these techniques, we aim to enhance suicide detection and prevention efforts in the context of the widespread use of social networking media.