A machine learning-based approach for data analysis to ascertain suicidal individuals from Social media users

Abstract

In this research, we propose a hybrid model for predicting suicide risk from text data that incorporates BERT, VADER, and a Random Forest classifier for sentiment analysis. This model aims to identify individuals who may be at risk of committing suicide based on the tone of the text. The model is trained on a labelled dataset of text data that is either classified as ”suicide” or ”not suicide,” which provides the model with instances of text data that are linked with high or low suicide risk respectively. In order to extract feature representations of the text data, the BERT model is utilized, and the VADER model is utilized in order to extract sentiment ratings for each individual text. These features are integrated into a single feature vector for each text, and then the Random Forest classifier is trained using this feature vector. A number of different metrics, including accuracy, precision, recall, and F1-score, are utilized in order to assess the performance of the model. The findings of this research indicate that the hybrid model that was suggested is capable of accurately predicting the risk of suicide based on text data and that it is suitable for use as a tool to help clinical decision-making. The performance of the model to recognize patterns and trends in text data that are indicative of suicide risk holds promise for future research in the subject. Our novel composite model combining BERT, VADER with Random Forest Classifier has the accuracy of 82 percent.