Performance Evaluation of SAE-O-BiLSTM for Predicting the Suicidal Behavior

Adolescents experience a complex transition from childhood to adulthood, focusing on the development of physiological and psychological factors. The various challenges associated with them include identity struggles, academic pressure and social adjustments, emotional adjustments etc. which contribute to serious mental health issues. If not addressed properly may lead to increase in the feelings of hopelessness and despair. Sometimes this kind of situation may lead to serious and complicated issues like suicide. Early and appropriate support can significantly reduce the risk of suicide among teenagers. Digital platforms and AI have the potential to enhance mental health and suicide detection and prevention and treatment outcomes. The study predicts suicide by using SAE-O-BiLSTM (Author’s model). The model makes use of the NLP and various deep learning algorithms to predict the suicide behavior in adolescents. Further the results are compared with the machine learning algorithm like Logistic Regression, Support Vector Machine and Decision Tree for the best suicidal prediction. The study clearly shows that the prediction of suicidal behavior is better predicted by the author’s model. The performance parameters taken into consideration are Accuracy, Recall, Precision and F1 Score.