Detecting Suicide and Self-Harm Discussions Among Opioid Substance Users on Instagram Using Machine Learning

Area of Science:

• Digital Health
• Public Health
• Social Media Analysis

Background:

• Suicide and substance use disorder (SUD) are critical public health issues for young adults in the U.S.
• Social media platforms are increasingly used by young adults, offering potential for public health surveillance.
• Leveraging social media can help detect suicide-related topics among individuals with SUD.

Purpose of the Study:

• To identify and characterize conversations related to suicide and self-harm on Instagram.
• To analyze the co-occurrence of these conversations with substance use disorder (SUD) discussions.
• To understand the themes within these intersecting online discussions.

Main Methods:

• Utilized big data and machine learning to collect and classify Instagram posts using 632 substance-related hashtags.
• Filtered posts for online drug diversion and subsequently for suicide/mental health discussions.
• Manually annotated posts for SUD and mental health themes, using Chi-square tests to analyze associations.

Main Results:

• Identified 719 Instagram posts/comments discussing suicide, substance use, and/or mental health.
• Posts mentioning SUD and mental health were significantly more likely to include suicide discussions (p < 0.001).
• Key themes included concurrent discussions of suicidal ideation, suicide attempts, and low self-esteem.

Conclusions:

• Preliminary evidence suggests a significant co-occurrence of suicide and mental health discussions with substance use disorder (SUD) on social media.
• This intersection represents a critical health risk factor, particularly on platforms frequented by young adults.
• Further research into specific ideation patterns is needed for developing digital suicide prevention strategies.