Utilizing Natural Language Processing to identify Suicide Attempt and Ideation in Text
Primary Investigator (PI) Name
Md Abdullah Al Hafiz Khan
Department
CCSE - Computer Science
Abstract
Amidst a rising mental health crisis, accurately identifying signs of suicide attempts and ideation in emergency communications remains a critical challenge. The purpose of this research is to leverage the capabilities of Natural Language Processing (NLP), particularly the Prodigy annotation tool, to create and refine classification models capable of detecting signs of suicide risk within 911 transcripts. The methodology involves: 1) utilizing data from a cleaned text file comprising five thousand records; 2) annotating in Prodigy, an interactive NLP tool that will be used to identify key patterns indicative of suicide involvement; 3) upon creating a comprehensive, annotated dataset, the final step involves training machine learning models to learn from the intricacies of the annotated dataset. The anticipated outcome of this research is the creation of a highly robust and sensitive classification model, thoroughly annotated using Prodigy, which is expected to significantly improve the detection of suicide risks.
Disciplines
Computer Engineering
Utilizing Natural Language Processing to identify Suicide Attempt and Ideation in Text
Amidst a rising mental health crisis, accurately identifying signs of suicide attempts and ideation in emergency communications remains a critical challenge. The purpose of this research is to leverage the capabilities of Natural Language Processing (NLP), particularly the Prodigy annotation tool, to create and refine classification models capable of detecting signs of suicide risk within 911 transcripts. The methodology involves: 1) utilizing data from a cleaned text file comprising five thousand records; 2) annotating in Prodigy, an interactive NLP tool that will be used to identify key patterns indicative of suicide involvement; 3) upon creating a comprehensive, annotated dataset, the final step involves training machine learning models to learn from the intricacies of the annotated dataset. The anticipated outcome of this research is the creation of a highly robust and sensitive classification model, thoroughly annotated using Prodigy, which is expected to significantly improve the detection of suicide risks.