Suicidal Ideation Detection

Suicide remains a critical public health concern, claiming over 700,000 lives annually worldwide. Traditional methods of suicide risk detection such as interviews and questionnaires often fall short due to underreporting, stigma, and the lack of real-time analysis. This study proposes a machine learning-based system to classify individuals as "at-risk" or "not at-risk" by analyzing psychological, behavioral, and demographic data. Models including Logistic Regression, Random Forest, AdaBoost, Gaussian Naïve Bayes, Decision Tree and Gradient Boosting are employed to identify hidden patterns in user input, enabling early and accurate prediction. The system addresses limitations of conventional approaches by automating risk assessment, reducing subjective bias, and offering scalable, real-time intervention capability. Data preprocessing techniques like sentiment analysis and feature encoding further enhance model accuracy and responsiveness. This research not only advances academic understanding of mental health prediction but also aims to make a tangible impact by equipping professionals with timely, explainable, and actionable insights to prevent suicide.

Keywords: Suicide Prediction, Machine Learning, Mental Health, Logistic Regression, Random Forest, AdaBoost, Gaussian Naïve Bayes, Decision Tree, Gradient Boosting, Sentiment Analysis, Early Intervention, Risk Assessment.