Suicidal Ideation Detection from Social Media Texts Using an Interpretable Hybrid Model

Suicide ranks as the second leading cause of death among adolescents worldwide, with suicide attempts associated with higher future risks of suicidal tendencies. While some with suicidal thoughts conceal them, many express their intentions on social media. In this paper, we have proposed an hybrid deep learning (DL) technique using Convolutional Neural Network (CNN) and Gated Recurrent Unit (GRU) for suicidal ideation detection from social network (SN) data. Also, we have applied Explainable AI (XAI) to briefly understand the outcome of the prediction and check the correctness of the model. A publicly available dataset has been used to train and test the model. To evaluate the model, several evaluation metrics has been utilized. Experiments have shown that the proposed model achieved 97.27% accuracy. We have compared the proposed model with several other existing Machine Learning (ML) and DL techniques and other existing literature. The proposed model has outperformed them all.