Stop Soldier Suicide Uses Pariveda Data Platform on AWS to Better Predict Risk for Veterans

Every year since 2001, more than 6,000 US military veterans die by suicide, with many of those showing no signs of mental illness or risk of suicide. Stop Soldier Suicide is on a mission to help stop this growing problem. The only national nonprofit focused solely on solving the issue of suicide among US veterans and service members, Stop Soldier Suicide provides evidence-based therapies and suicide-specific interventions for the most at-risk veterans and service members. “Our vision is a nation where veterans and service members have no greater risk for suicide than any other American. We have an aggressive goal to reduce the military suicide rate by 40 percent by no later than 2030,” says Austin Grimes, chief product officer at Stop Soldier Suicide. To support its mission, Stop Soldier Suicide needed to understand its data better to see trends and offer new insights to those who are helping the veteran community prevent the veteran suicide rate from rising.

As part of its efforts, the organization started the Black Box Project, a forensic data analysis project that uses artificial intelligence to identify and analyze clinically meaningful data from digital devices of veterans who died by suicide. However, Stop Soldier Suicide struggled to analyze the data collected from the project. Grimes says, “We were stuck with local, disorganized data that simply could not be used due to the amount of variance in both data quality and taxonomy of extracted device data.” To solve the problem, Stop Soldier Suicide wanted to create a new solution that quickly ingested data from different sources and simplified analysis. “We wanted to build the world's best suite of tools and models that can understand and predict an individual’s risk and build a deeper understanding of who these veterans and service members are,” Grimes says.

As an Amazon Web Services (AWS) customer, Stop Soldier Suicide knew it wanted to build a cloud-based solution. The organization met that requirement by working with Pariveda, a technology and management consulting firm and AWS Partner. Grimes explains, “Because we are a lean nonprofit organization without internal capabilities to fully build and deploy cloud infrastructure, Pariveda serves as both an engineering team and strategic advisor.”

Pariveda helped build a Suicide Intelligence Platform (SIP) to ingest various data sources to aid in predicting and preventing death by suicide of veterans and active duty military personnel. “We wanted to create a platform that brings in additional clinical data to enrich what’s already collected from Black Box Project,” says Daniel Clements, manager at Pariveda. "With SIP, we can help Stop Soldier Suicide produce unique insights into the causes and correlating factors of suicide in the veteran and service member population.” In addition to building the platform’s architecture, Pariveda led a multi-day workshop with Stop Soldier Suicide leaders to create the platform’s vision and find ways to bring data-driven clinical experience into practice through predictive data analysis.

The SIP solution ingests Black Box Project and health cloud data and stores it in Amazon Simple Storage Service (Amazon S3). The solution relies on AWS Glue to index data, with governance through AWS Lake Formation. Data enrichment and transformation between layers is managed via Apache Spark, using both AWS Lambda and AWS Glue. Stop Soldier Suicide data scientists interact with data from the platform by taking advantage of Amazon QuickSight dashboards for visualization and exploration, plus Amazon SageMaker studio notebooks for analytics and machine learning. “The other piece that’s fundamentally brought about by the architecture of this platform is the ability to enrich and transform data at great scale on AWS,” Clements says. The SIP platform went live in March 2024.

With the SIP solution, Stop Soldier Suicide can quickly ingest and enrich device data with health data and other contextual information. Ingestion takes place in minutes, and Stop Soldier Suicide can redact and cleanse millions of data points per device in hours. The organization can import this data from other relevant sources at the click of a button. Grimes says, “Through our work with AWS and Pariveda, we can now ingest, normalize, and redact device extractions on the fly, which, in turn, allows our data scientists and engineers to focus on working with the data as opposed to cleansing and prepping.”

With a broader repository of knowledge, Stop Soldier Suicide is increasing the quality and depth of its analysis of what contributes to the risk of suicide. For instance, the organization is conducting machine learning on the data, along with known suicide risk factors such as social isolation. “By analyzing things like frequency of contacts and conversations, we now have ways to objectively measure these risk factors,” says Grimes. This is helping the organization uncover new insights and make better predictions. “Through a focus on predicting acute levels of risk with the Pariveda and AWS platform, we can assist clinicians in applying the right resources at the right time to mitigate the risk of suicide,” Grimes says.

Stop Soldier Suicide is continuing its work with Pariveda, building on the foundation the two organizations have built together. Grimes says, “We anticipate having predictive models in the hands of clinicians in early 2025, which will be a major breakthrough for Black Box Project and the Suicide Intelligence Platform.”