Using NLP-based RCNN to Detect Suicide Indicators in Social Media Posts: A Proactive Approach to Lowering Suicide Rates

Disciplines

Artificial Intelligence and Robotics | Data Science | Other Mental and Social Health | Quantitative Psychology

Abstract (300 words maximum)

Suicide has become a significant issue within our society over the past few decades, so much so that 13.42% of 100,000 people will commit suicide within the year. When looked at on a larger scale, this is a drastic amount of people dying, which could be helped or stopped. This is shown by the percentage of a 25% lower suicide rate after having voluntary talk therapy. With an issue that can be helped by talking to a trained professional, it is better to find the person before they have gone too far. This research that we are conducting will help create a proactive way to help determine people for depression. A proactive approach will help lower suicide rates and give medical professionals a better chance at helping people in need before there is a chance to commit suicide. There have been leaps and bounds in machine learning in recent years, and using different databases, we can train an ML model with Natural Language Processing (NLP) techniques to determine if someone is considering suicide. We can train it using a 500 anonymized Reddit post dataset that has all the posts labeled based on the post’s relation to suicide. In this work, we propose to develop an NLP-based Recurrent Convolutional Neural Network (RCNN) to detect suicidal events such as suicide attempts and ideation. Using available word embeddings, we will represent Reddit posts as a feature vector and feed these features to an RCNN network to detect suicidal events. We envision comparing our model performance with traditional machine learning algorithms, such as Logistic Regression, Feed Forward Neural Networks, etc., to showcase the effectiveness of our algorithm.

Keywords: Mental Health, Suicide, NLP, Recurrent Convolutional Neural Network, Feed Forward Neural Network

Academic department under which the project should be listed

CCSE - Computer Science

Primary Investigator (PI) Name

Md Abdullah Al Hafiz Khan

This document is currently not available here.

Share

COinS
 

Using NLP-based RCNN to Detect Suicide Indicators in Social Media Posts: A Proactive Approach to Lowering Suicide Rates

Suicide has become a significant issue within our society over the past few decades, so much so that 13.42% of 100,000 people will commit suicide within the year. When looked at on a larger scale, this is a drastic amount of people dying, which could be helped or stopped. This is shown by the percentage of a 25% lower suicide rate after having voluntary talk therapy. With an issue that can be helped by talking to a trained professional, it is better to find the person before they have gone too far. This research that we are conducting will help create a proactive way to help determine people for depression. A proactive approach will help lower suicide rates and give medical professionals a better chance at helping people in need before there is a chance to commit suicide. There have been leaps and bounds in machine learning in recent years, and using different databases, we can train an ML model with Natural Language Processing (NLP) techniques to determine if someone is considering suicide. We can train it using a 500 anonymized Reddit post dataset that has all the posts labeled based on the post’s relation to suicide. In this work, we propose to develop an NLP-based Recurrent Convolutional Neural Network (RCNN) to detect suicidal events such as suicide attempts and ideation. Using available word embeddings, we will represent Reddit posts as a feature vector and feed these features to an RCNN network to detect suicidal events. We envision comparing our model performance with traditional machine learning algorithms, such as Logistic Regression, Feed Forward Neural Networks, etc., to showcase the effectiveness of our algorithm.

Keywords: Mental Health, Suicide, NLP, Recurrent Convolutional Neural Network, Feed Forward Neural Network