Black Box Project, our winning solution, is a pioneering innovation that leverages artificial intelligence to redefine our understanding of suicide risk among Veterans.
We have yet to truly understand the private, behind-the-walls details of Veterans’ lives during the last minutes, hours, days, and weeks before they die by suicide.
Epidemiological data are strong, but they don’t give us enough information about immediate individual risk. When we try to understand individual risk retrospectively, current psychological autopsies represent the limited perspective of those around the Veteran.
To date, we have never been able to capture the inner thoughts, feelings, and behaviors of those who died by suicide in the time immediately leading up to the event.
Without this individual-level, real-world data, current prevention efforts will continue to fall short of the ability to save more Veterans at risk.
Today, we share more intimate information with our digital devices than with anyone else in our lives. Our devices capture what we say, do and how we interact in both the digital and physical worlds.
At Stop Soldier Suicide, we believe this information can speak volumes about suicide risk and may uncover acute details in the days, weeks, and months before a death.
Modeled after the “black box” flight recorder used in aviation, our solution, Black Box Project, will conduct digital autopsies for Veterans who have died by suicide to recreate the final moments of life.
We’ll uncover never-before-known insights to completely redefine the way we understand signals of risk, and we’ll advance methods of outreach and care for Veterans at highest probability for suicide.
We leverage a diverse suite of Amazon Web Services (AWS) tools to unlock the potential of Black Box Project. Products such as Amazon Simple Storage Service (S3), AWS Athena, AWS SageMaker, and AWS Comprehend turn potentially disparate extraction and analysis processes into a streamlined tool with a consistent technology profile.
Machine learning algorithms, natural language processing, and entity extraction techniques – powered by AWS – are used to build models of pre-suicidal behaviors highly correlated with suicide, which uncover novel signals and insights that can be shared with the veteran-serving community to save lives at scale.