Suicide Ideation Detection: Harnessing Machine and Deep Learning for Early Risk Identification

The prevalence of suicide and depression has become a growing concern worldwide. Early detection of individuals at risk plays a crucial role in providing timely interventions and support. In this study, we explore the effectiveness of various machine and deep learning methods for the task of suicide and depression detection using the Suicide and Depression Detection dataset obtained from Kaggle. The dataset comprises 232,074 unique instances with two target classes. Through extensive experimentation, we observed that Support Vector Machines (SVM) achieved the highest accuracy of 95%, closely followed by Logistic Regression with an accuracy of 93.96%. Additionally, Random Forest, Passive Aggressive Classifier (PAC), and Long Short-Term Memory (LSTM) demonstrated competitive performances, achieving accuracies of 93%, 93%, and 92.84%, respectively. 1D Convolutional Neural Network (CNN) yielded an accuracy of 92.73%. Our findings highlight the potential of these models in identifying individuals at risk of suicide and depression, contributing towards proactive mental health care initiatives. Furthermore, the combination of 1D CNN and LSTM achieved an accuracy of 92.26%. These results emphasize the effectiveness of these models and their potential impact on proactive mental health care initiatives.