How AI and ML can help prevent veteran suicides
Veteran suicide is one of the most pressing issues for the veteran community today. Although the rate of veteran suicides has fluctuated in the past few years, the 2022 National Veteran Suicide Prevention Annual Report from the U.S. Department of Veterans Affairs (VA) states that in 2020, suicide was the second most common cause of death among veterans under the age of 45. The report also shows that about 17 veterans committed suicide every day in 2020.
The U.S. Department of Defense (U.S. DoD) and VA offices are tackling this ongoing problem by increasing support services to veterans and researching innovative mitigation methods. One revolutionary solution they are looking into is the use of predictive modeling and machine learning (ML) to prevent these tragedies by identifying suicidal ideation earlier.
Veteran suicide is a significant problem in our country today. Suicide is the second leading cause of death for veterans under the age of 45. Veterans are at a 57% higher risk of suicide than nonveterans. And the problem is getting worse. Over the last two decades, veteran suicide rates have risen by 50% to 90%, depending on age group.
We can cite yearly statics on suicide rates per 100,000 veterans. However, if you are a veteran who is affected by suicidal ideation or suicide, or that veteran’s family member or friend, this is a big problem regardless of statistics. As a nation, we can’t let our veterans down.
The U.S. Department of Defense and Department of Veterans Affairs are implementing initiatives to help prevent veteran suicide. The VA has been proactive in funding research to identify causes and detect potential suicide risks early. There are dedicated researchers at many VA facilities, and universities and research consortia are funded to address possible solutions.
In addition, the VA has invested in IT infrastructure to facilitate and encourage research and experimental development. They are also funding a study to explore the best practices and innovative technologies available from industry partners that can be used to help prevent veteran suicide. And finally, the VA has implemented a veteran suicide hotline to make it easier to connect those at risk with someone who can help quickly.
No technology can fully replace clinical care, proactive counseling, and the concern of friends and family. However, there are ways in which technology can be used to reduce the staffing required to identify suicidal risks and to spot these signs much earlier.
Letting problems fester and amplify undetected makes them harder to address. ML techniques can flag high-risk cases quickly so that clinical staff and counselors can use their limited resources to address the highest-risk patients first, and to reduce the chances of a patient flying under the radar.
The DoD is using models in research facilities to spot early signs of PTSD, suicidal ideation, and risky behaviors in soldiers. The VA can use these same technologies to help identify behaviors in veterans that might indicate when they are at high risk for suicide. Also, the right IT architecture at the data layer can make it less expensive and less cumbersome to share insightful datasets across facilities within and outside the VA—all while protecting private patient data.
The VA has already implemented programs and funded projects with industry partners and research consortia to create models to detect risk. The challenges are to thoroughly test the models and to establish a high level of confidence among counselors and medical professionals. Meeting these challenges is necessary to ensure that the models can balance false positives and negatives while not overfitting models based on narrow patient populations.
There are many challenges and roadblocks that prevent the VA and DoD from using AI. The first challenge is to make sure that the models are applicable across the entire population of veterans. The next challenge is to reach a consensus within the medical community about the right balance of filtering out false positives at the expense of accidentally generating false negatives.
The final challenge is the speed and scale necessary to keep these models updated. As populations and conditions fluctuate, these models have to be continually updated. This level of accuracy and size of the model libraries requires extensive IT resources, data engineers, and data scientists to keep a large model library updated and fresh.
NetApp® solutions can help address these challenges by reducing IT complexity and speeding up the workflow and data pipeline to keep models updated and accurate. Additionally, we can reduce the challenges in bridging data across facilities that may be either inside or outside of the VA medical system. The VA’s data scientists, clinicians, and subject matter experts are already working at capacity.
We can reduce the IT burden by creating a seamless experience to share datasets and models among researchers quickly and to reduce the burden to operationalize and keep models relevant.
Today, if a scientist wants to share a large data model, that sharing can take hours or days, depending on the size of the dataset. That same scientist could share a Google doc almost instantly. NetApp tools can make sharing models more like sharing Google Docs, increasing the speed and reducing the time to keep the model updated. This also reduces schedule risk for projects.
We look forward to working with our technology partners to reduce the VA’s IT burden so the experts can create high-fidelity production models that clinicians and counselors can use as decision aids to do their job. All while keeping data, privacy, and HIPPA safeguards in place.
Sunjit currently leads our senior team of business developers across DoD/IC/Fed-Civ/SLED and serves as chief of strategy for the NetApp public sector. He has 20+ years in the technology industry ranging from startup companies in the financial domain to large systems integrators: L3Harris, Northrop Grumman, and Peraton. At Northrop Grumman, he worked in account management, BD, capture, and program management. He also led the AI/ML capture center of excellence as an industry tech fellow. At Peraton, he was the CTO/Director of Technology for one of the four business sectors. His technology background includes data science, systems engineering, aviation, telecommunications, C4ISR, and genetics.
Brush up on the latest trends and developments in cloud, on premises, and everywhere in between. This is where it all gets real, with a cherry on top.
Explore a wide range of open forums where you can post questions, share answers and just generally get smart on all the NetApp technologies that matter most to you.