Development of Internet suicide message identification and the Monitoring-Tracking-Rescuing model in Taiwan.

Area of Science:

• Computational psychiatry
• Digital mental health
• Artificial intelligence in healthcare

Background:

• The internet facilitates the rapid spread of suicide-related content, highlighting its critical role in suicide prevention.
• Identifying online suicide risks through advanced technology can aid in predicting and intervening in suicidal behavior.

Purpose of the Study:

• To develop and evaluate an artificial intelligence (AI) model for detecting suicidal ideation in online social media posts.
• To establish a Monitoring-Tracking-Rescuing model integrating AI-driven risk assessment with professional validation for suicide prevention.

Main Methods:

• Utilized text mining and Natural Language Processing (NLP) techniques, including Long Short-Term Memory (LSTM) and Bidirectional Encoder Representations from Transformers (BERT), on Taiwanese social media data.
• Implemented a two-step high-risk identification process: AI-based classification of suicide risk followed by validation by mental health professionals.

Main Results:

• The AI model achieved 80% sensitivity and specificity in identifying high-risk suicide content from a dataset of 404 high-risk and 2226 low-risk posts.
• The developed model successfully classified online messages related to suicidal ideation or behavior.

Conclusions:

• AI techniques can effectively detect and monitor high-risk suicide posts, alerting mental health professionals.
• Periodic tracking combined with manual validation is recommended to improve the reliability and effectiveness of online suicide prevention strategies.