Machine Learning-Based Suicide Risk Prediction Model for Suicidal Trajectory on Social Media Following Suicidal...

Area of Science:

• Computational psychiatry
• Social media analytics
• Mental health technology

Background:

• Machine learning and natural language processing show potential for predicting suicide risk from social media.
• Previous research indicates the feasibility of using AI for mental health monitoring.

Purpose of the Study:

• To externally validate the Suicide Artificial Intelligence Prediction Heuristic (SAIPH) algorithm using independent cohort survey data.
• To assess SAIPH's efficacy in tracking changes in suicidal ideation (SI) over time.
• To explore factors influencing suicidal trajectories on social media.

Main Methods:

• Twitter (X) data from student and COVID-19 cohorts were scored using SAIPH.
• SAIPH scores were compared against validated survey instruments for depression and SI.
• Longitudinal analysis of SAIPH scores was conducted to evaluate SI trajectory changes.

Main Results:

• SAIPH scores interacted with posting frequency to correlate with quantitative SI metrics in both cohorts.
• The daily SAIPH score slope predicted changes in SI over short periods (≤2 weeks).
• SAIPH identified a negative association between SAIPH score and time in individuals with suicidal mentions, particularly within 72 hours of response.

Conclusions:

• SAIPH is validated as a measure associated with perceived stress, SI, and SI changes.
• The study introduces novel methods for assessing social media's impact on suicidal trajectories.
• SAIPH offers a tool for real-time monitoring and understanding of mental health trends.