Suicide Ideation and Detection

Project Title

Suicide Ideation and Detection

Project Area of Specialization Artificial IntelligenceProject Summary

Suicidal ideations (SI), often called suicidal thoughts or ideas, is a broad term used to describe a range of contemplations, wishes, and preoccupations with death and suicide. Suicide remains the second leading cause of death among the young generation with an overall suicide rate of 10.5 per 100,000 people. Sometimes people post more personal content that shows signs of something going wrong in their life. It’s a call for help, these posts may have words to convey fear, loneliness, and hopelessness yet we don’t even realize it or know for sure.

This study sought to analyze “Suicide Ideation” documents posted on social media with an aim to determine reasons for suicide actualization and to analyze the post to know behind the meaning of that statement which if a person is going to suicide or not. For that purpose, we will use NLP to do sentimental analysis and use machine learning and some advanced technique like transformer models to train our model so that we can accurately predict if a person is going to do suicide or not. On top of that, we are building a Chatbot for front-end development that will predict whether the conversation is suicidal or not accurately and provide help, awareness, and give suggestions to improve the mood of a person.

Project Objectives

• The goal of this project is to build a model that classifies social media posts as suicidal and not suicidal accurately (for research work).
• Integrates it with front-end technology to build a chatbot that will tell if the conversation of a person with a chatbot is suicidal or not.
• To help, spread awareness, and give suggestions to improve the mood of a person.

Project Implementation Method

Data Collection

• Reddit Dataset

Preprocessing Data

• Exploring Dataset
• Cleaning Dataset
• Stop Word Removal
• Tokenization
• Stemming

Feature Extraction

• Bag of Word
• Term Frequency Inverse Frequency (TF-IDF)

Training Model

• Logistic Regression
• Decision Tree Classifier
• Random Forest
• Naive Bayes

AI Chatbot

• Based on the RASA framework

Benefits of the Project Problems 

• A person suffering from suicidal ideation isolates himself/herself from society and rarely discloses problems to anyone.
• The person tends to ignore the illness and doesn't consult a therapist/psychiatrist. The reason could be many but often a person doesn't want to share because of trust issues or want to void being judged.

Benefit

• The chatbot will provide feedback reports by performing sentiment analysis and will suggest if some suicidal sentiment is found.
• It's a non-judgemental AI chatbot, so any depressed person who may be insecure about asking help can chat here easily.
• It will also provide help by suggesting some doctors and psychiatrists.
• It will also help to improve the mood of a person.

Technical Details of Final Deliverable Research Work on Dataset

We will be using a labeled dataset collected from the subreddit of the Reddit social platform and did research on this dataset to find the accuracy using different models mentioned project implementation section.

For Chatbot

For chatbot, we have applied the support machine algorithm (SVM). 

With the help of a trained model, we will create chatbot that will understand our input and prove brief conversation to ask users how they are feeling and what's going in their life.

RASA

We have used RASA for our chatbot. RASA is an open framework for natural language understanding, dialogue management, and integration.

Chatbot Intent 

Intent and their example are used as training data for the assistant natural language understanding (NLU) model.

We group these examples according to the idea of the goal of the message is expressed.

Chatbot Stories

Stories are used to train our assistant's dialogue management model.

Stories can be sued to train models that are able to generalize to unseen conversation paths.

It is a representation of a conversation between a user and an AI assistant.

Final Deliverable of the Project HW/SW integrated systemCore Industry HealthOther Industries Medical , Others Core Technology Artificial Intelligence(AI)Other Technologies OthersSustainable Development Goals Good Health and Well-Being for PeopleRequired Resources

Item Name	Type	No. of Units	Per Unit Cost (in Rs)	Total (in Rs)
			Total in (Rs)	35000
System Requirements	Equipment	1	35000	35000