Appendix A

Statement of Task

A planning committee of the National Academies of Sciences, Engineering, and Medicine (the National Academies) shall plan and host a 2-day public workshop to explore the current scope of activities, opportunities, benefits, and risks of leveraging real-time data sources and innovative data science techniques and methods, including artificial intelligence/machine learning (AI/ML) algorithms, to help inform suicide prevention efforts at the individual, community, and population levels.

The workshop will feature invited presentations and moderated discussions on topics that may include:

• Valid, real-time data sources, including deidentified population-level data, to assist with early warning and to identify potential hotspots for fatal and nonfatal suicide-related outcomes.
• Innovative methodologies, including AI/ML algorithms that can be leveraged to identify individuals, groups, communities, and populations at high risk for suicide.
• Innovative data science techniques and methods, including AI/ML algorithms, to identify, predict, and refer individuals at risk for suicide to appropriate care and services, using tools such as
  • suicide risk prediction algorithm methodologies used by technology and social media platforms to identify users at risk for suicide,
  • relative effectiveness of social suicide prediction algorithms in accurately identifying individuals at risk compared to medical suicide prediction, and

• • algorithm updates in the context of the “988” mental health crisis hotline, to be launched in July 2022.
• Strategies to match education/referrals provided with the risk stratification of the AI/ML suicide prediction algorithm, including
  • a comparison to evidence-based approaches/best practices for mental health/behavioral health crisis response.
• Additional opportunities/gaps such as
  • evidence-based/best practices for online peer support groups,
  • considerations for individuals in rural/underserved areas without broadband access or limited access to the internet, and
  • potential opportunities for follow-up after identification by social suicide prediction algorithms, while ensuring privacy.
• Potential risks, unintended consequences and pitfalls of leveraging AI/ML algorithms for identifying individuals at risk for suicide or experiencing a behavioral health crisis.
• Evidence-based approaches/best practices to optimize benefits and minimize harm/unintended consequences of suicide prediction algorithms.
• How best to connect these innovative data science approaches with ongoing suicide prevention efforts in communities and health systems.
• Evidence /research /program evaluation gaps to measure effectiveness/efficacy of suicide prediction algorithms at the individual, community, and population levels.
• Next steps and potential opportunities for action to support upstream suicide prevention efforts.

The planning committee will develop the agenda for the workshop sessions, select and invite speakers and discussants, and moderate the discussions. A proceedings of the presentations and discussions at the workshop will be prepared by a designated rapporteur in accordance with institutional guidelines.