Office of Research & Development
The NAII fosters collaboration across disciplines and promotes sharing of ideas and resources to create AI solutions that help Veterans. We build strategic partnerships across the private sector, the federal government, and the research community to maximize AI's potential.
Below are some of the high-impact pilot projects that NAII has helped spearhead and develop.
The COVID-19 Mortality Model provides VA medical staff with insight to help them predict the severity of a patient's disease course and plan patient care accordingly. © iStock/NataBene
Early in the pandemic, the NAII collaborated with the Washington DC Veterans Affairs Medical Center (VAMC) to provide resources and guidance in developing a machine learning model to predict the severity of a COVID-19 patient’s disease course. The model helps medical staff identify high-risk patients who are isolated at home. The model includes post-hospitalization insight that can help clinicians determine how to allocate COVID treatment to acute COVID patients. The model aggregates multiple data sources to support clinicians’ decision-making.
The NAII has shared the COVID-19 mortality model with 12 VA urban and rural pilot sites. The model helps VA determine how AI can support operations, strategic planning, and palliative care coordination. In 2021, the COVID-19 mortality model transitioned to broader use across VA and has been piloted at 13 VAMCs nationally. Additionally, the Kansas City VAMC adapted the model to assist clinicians in setting up appropriate levels of post-discharge care. A minimally viable product is currently in development; early users will provide feedback for future product development.
The AI-To-Go Tool allows researchers and practitioners to track AI models from validation to user acceptance and eventually enterprise-level implementation.© iStock/metamorworks
The AI-To-Go Tool is a cloud-based pipeline for AI research and statistical models. The tool allows researchers and practitioners to access and track AI models from validation to user acceptance and eventually enterprise-level implementation. AI-To-Go will enable models to be stacked and compared against one another for performance while receiving input from clinicians on the value and impact of the information presented. If successful, the model will be transitioned to a pilot.
NAII Research Helps Promote Trustworthy AI
The AI-To-Go Tool and COVID-19 Prognostics Explainable AI Tool were developed to promote the Principles for Use of AI in Government (Sec. 3) outlined in Executive Order 13960: Promoting the Use of Trustworthy Artificial Intelligence in Federal Government. The NAII is committed to developing AI tools that are purposeful and performance-driven, transparent, accountable, regularly monitored, respect our nation's values, and remain consistent with the Constitution by protecting privacy, civil rights, and civil liberties.
Once fully operationalized, the Digital Command Center will use predictive analytics and AI to make it easier for VA to track and share data, thereby improving coordination of patient care.
The Digital Command Center (DCC) aims to improve the health and well-being of Veterans by leveraging the VA's AI capabilities, infrastructure, and expertise. Once fully operationalized, the DCC will use predictive analytics paired with AI to make it easier for VA to consolidate, track, and share data among medical center administrators.
The data consolidation and centralization at DCC will enable VAMCs to improve operational efficiency, bed space management, quality of care, patient experience, clinician satisfaction, transparency, coordination, and staffing. As an AI-powered digital hub, the DCC will remove delays and enhance timely decisions for Veterans' benefit.
The Veterans Crisis Line will more quickly identify and help Veterans in crisis, if the current pilot proves successful. © iStock/Nicola Katie
When Veterans are in crisis, every moment counts. The NAII supports SoKat, a consulting firm working on AI solutions, to pilot an AI-powered suicide ideation engine. The pilot will determine if a new Natural Language Processing (NLP) engine will enable quicker, more accurate filtering of text responses. If so, the Veterans Crisis Line will be able to more quickly identify and help Veterans in crisis.
The SoKat-developed suicidal ideation detection engine will help VA detect, evaluate, and report indicators of suicidal ideation, self-harm, emotional dysregulation, pain, and other mental health issues and report results.
The VA is using artificial intelligence to improve the capabilities of smart wearable products.
© iStock/Edwin Tan
Wearable technology has rapidly improved in recent years. Through advances in vital signs monitoring − such as for oxygen, heart, and sleep − AI-powered technologies can now diagnose patients and detect abnormalities. Through continuous sensing and monitoring, wearable devices provide health care professionals with valuable information about a patient’s health.
A joint effort by VA and the Snyder Lab at Stanford University, with collaboration from the Department of Defense (which will expand its efforts into the Department of Homeland Security), the Precision Brain Health Pilot uses live data from a combination of wearables, (multiomic) sampling, and additional data stores to enhance monitoring capability.
AI is used to analyze the combination of data to (1) develop and test methods for better post-traumatic stress disorder (PTSD) and suicidality detection and treatment, and (2) track the effects of stress, sleep, diet, and other variables from basic training through a Veteran’s retirement.
Using cutting-edge analytics with novel secure data exchange and sharing methods, the Precision Brain Health Pilot will help provide a complete view of a patient’s health, while contributing to advanced care and knowledge of the human body.
Are you a VA investigator interested in collaborating with the NAII to pilot cutting-edge, high-impact AI research? Contact us at NAII@va.gov.