MERIT Delivers on Its Name with AI to Improve Military Medical Readiness

Our nation’s warfighting ability depends on many things. One of the most important: the health of U.S. service members. That’s why medical fitness is a top military priority. When personnel can’t deploy, the effects ripple far beyond service-level readiness. For example, Department of Defense pays an estimated $3 billion annually in personnel costs for non-deployable Army soldiers alone.

Service members who can’t deploy can find themselves in a state of extended limbo: unable to perform their duties, unable to leave the military, and unable to collect compensation until their case is resolved—a process that can take years.

And then there’s the bigger human toll. In the past five years, post-traumatic stress disorder remains the #1 cause of service member disability. (Other drivers include major depressive disorders and traumatic brain injuries.) Such “invisible wounds” affect not only DoD readiness but individual wellbeing.

Left untreated, these conditions can lead to devastating outcomes. Approximately 17 veterans succumb to suicide each day.

The period before referral to the military’s Disability Evaluation System, or DES, provides a crucial opportunity to intervene early and assist at-risk service members. That’s why the Defense Health Agency (DHA) engaged MITRE—operator of federal R&D centers for defense, health, and enterprise modernization—to study how existing health data sources could be used to help predict DES referral.

From this study, we developed the AI-driven MERIT, short for Medical Evaluation Readiness Information Toolset. We anticipate the capability will decrease the pre-DES period by 180 days—and potentially save billions of dollars.

Retired Col. William O’Brien, former DoD-appointed highly qualified expert in disability policy, originally advised the department on applying AI to improve the disability-determination process. “MERIT presents an opportunity to transform how DoD will handle medical readiness and a service member’s journey through the Military Health System,” he says.

“The MITRE team isn’t just processing data and punching numbers. They’re developing systems and informing policies that will make a real difference in the lives of warfighters.” 

Tackling the Top-24 Drivers of Military Disability

In collaboration with DHA, we leveraged established DoD health platforms to advance MERIT’s predictive-modeling capabilities. Our team developed a unique ensemble machine-learning approach, applying targeted, timely analytics on the 24 most common career-ending disability conditions.

MITRE’s Rachel Armstrong, the project’s co-lead, brings perspective as a former Army nurse and retired colonel. “MERIT is now available to give DoD healthcare providers and unit commanders greater confidence in making really tough decisions about when to refer to DES,” she says.

The toolset allows clinicians to assess individuals through a “Medical Provider” interface, which acts much like a car’s “Check Engine” light. Practitioners can see the probability of disability trending over time. This can inform discussions around treatment options and DES referral decisions.

With readiness the prime driver, our team moved quickly to get the tool ready for prime time. MERIT germinated from the seed of an idea from AI engineer Huang Tang, who built on a concept he encountered during related research. The team then took the idea from a lunch discussion with senior DoD officials to an Army Minimum Viable Product, called MVP in the acquisition world, in just 28 months, an almost unprecedented timeline.

“DES referral is a complex decision that impacts a service member’s career and livelihood,” Tang says. “MERIT aims to identify whether someone’s health condition will deteriorate within the next six months to where they can no longer perform their current duties—and help get ahead of the problem.”

Holistic Approach to Move the Needle on Readiness

MERIT is poised to deliver big on that goal. Demonstrating robust results against all tested conditions, the patent-pending capability represents a first for this kind of AI application. A few prior studies used modeling to predict DES referral. But those were based on limited data, such as for insurance purposes. MERIT considers the whole of a service member’s medical history.

MITRE’s role as a trusted adviser facilitated the effort of providing DoD with not only better data but the infrastructure to power such analytics. Our team accessed data from electronic health records dating from 2000 to 2019—more than three million Army soldiers.

They also established the data journey to prove the data accessible, accurate, and vigorous enough to support this type of time-sensitive decision-making. As Armstrong explains, what started as a research prototype has evolved into a system that’s a potential game changer for military personnel.

“Originally we developed models based on a historical data set,” she says. “Our more recent work established a pipeline fed by current data. A provider can see results for a specific service member today.

“We think MERIT could enable the military to really move the needle on readiness.”

Looking ahead, the team is testing MERIT’s user interface within Army operational units to support further validation. Its applications could also improve treatment of chronic health conditions more broadly.

Moreover, Armstrong says, “Our research advances represent considerable first steps toward DHA becoming AI-ready in the next few years to significantly advance the state of DoD healthcare.”

Join our community of innovators, learners, knowledge-sharers, and risk takers. View our Job Openings.