Transcom pursues AI to enhance patient movement ops and mass casualty response

SCOTT AIR FORCE BASE, Ill. — MIT Lincoln Laboratory researchers are developing and demonstrating bespoke artificial intelligence and machine learning assets to enhance U.S. Transportation Command’s capacity to perform in contemporary operations, and ultimately prepare for future conflicts.

“Right now, [the lab is] working with Transcom’s Surgeon General on how to build analytic tools with patient data to handle a mass casualty event,” John DeLapp, the futures division chief at the command’s analytics center, or TCAC, told DefenseScoop.

Other projects include but are not limited to advanced AI for large-scale in-flight messaging analyses, and algorithms to inform global air refueling missions. They’re all unfolding via a well-established partnership that the command has dubbed the MIT Lincoln Lab Living Plan. 

Officials briefed DefenseScoop on this collaborative initiative during an exclusive tour of the command’s headquarters at Scott Air Force Base, Illinois, this week. They also shared new details about an upcoming exercise that aims to optimize data transmission across a broad spectrum of operations.

MASCOT

Launched by the command in 2011, the lab’s “living” plan was conceptualized to be more dynamic than typical static business initiatives — and to enable the two partners to adapt as technologies, circumstances and demands change over time.

“Back then, and even to an extent now, Transcom has had some of the same challenges — stovepiped systems and data, and getting it all together. So, the idea was for Lincoln to help us assess data architectures and to build flexible and robust planning tools, and analytics that would tie planning and analysis together, and propose cross-cutting solutions to that,” DeLapp explained. 

“And what’s happened over time is there have been projects where Lincoln has worked with our air component, Air Mobility Command, on predictive maintenance. They’ve worked with our J6 [command, control, communications and computer systems directorate] in the cyber area. They’ve worked with the J2 [intelligence directorate] in characterizing foreign influence using open-source information,” he said.

Lab officials don’t build and maintain the widgets or systems, but instead prototype technologies and then hand things off to the command or a vendor to fully produce and deploy them. 

“For example, in the air refueling area, our analysis models — with contested environments and all the new challenges — these models are being asked to do more. The model run times are getting slower. It’s getting harder to do them. So, we’ve brought Lincoln Labs on to go research and develop newer algorithmic approaches to speed up techniques and to speed up those algorithms. And then that will get handed off to the contractor that builds and updates it, and then they’ll take that and then incorporate it into the model,” DeLapp said.

In fiscal 2020, the command asked MIT researchers to conduct a comprehensive assessment to gauge which of its directorates were the most and least ready candidates to adopt AI and machine learning in tailor-made use cases.

“That was pretty telling because you can have all this technology, but if the directorate doesn’t know what data they have, or they don’t even know if that data exists, if they don’t even know what a tool can do and what it can’t do, or what it should do — then the time’s not right for you to just go and embrace that,” DeLapp noted.

Roots of the lab’s ongoing analytic and AI-enabling efforts for Transcom’s Surgeon General stem back to challenges pinpointed in that eye-opening assessment. Among its weighty, far-reaching responsibilities, the command serves as the Defense Department’s sole manager for global patient movement, which often involves the rapid and high-stakes transportation of ill and injured military personnel. 

The speed of conflict and urgency of such medical operations can make data-capture and processing typically more complex for Transcom in this context.

Spotlighting such challenges, Transcom public affairs official Erik Anthony told DefenseScoop: “There are some locations where we’ll get paper documentation for these individuals. So when you have even 10 patients moving — and then you do this at scale with thousands of patients — and you have medics running to an aircraft with papers, and then these papers have to make it to three other stops where they’re doing continuity of care.”

The patient movement mission set was identified during the lab assessment as potentially being well-suited for application of AI and ML technology. In fiscal 2021, MIT Lincoln Laboratory began working with the U.S. Transcom Command Surgeon to develop an AI-assisted routing tool to support that mission, he explained.

Now, that’s just one in a suite of tools the lab is developing and refining to facilitate the scalable handling of patient data, automate identification of key patient movements, and support planning for clinical and air assets.

“The Mass Casualty Operations Toolkit — MASCOT — uses [AI and ML] to enhance accuracy, improve workflows, and provide critical insights for both routine patient transport and mass casualty events,” Anthony said.

Once prototyped, the outputs can go on to be maintained as separate tools or used as a specification for incorporation into the Transcom Regulating and Command and Control Evacuation System.

According to Anthony, by applying natural language processing and large language modeling in their toolkit, the lab will assist the patient movement community “in reducing their cognitive load when processing patient movement requests resulting in reduced errors and faster decisions, allowing for increased velocity of patient movement operations.”

“This effort seeks to aid clinicians and airlift planners in decision-making, focusing on addressing operational needs for effective medical transport, particularly during a mass casualty response,” he said.

Another in-development tool in the MASCOT arsenal is what officials referred to as a synthetic patient generator.

“When command officials do exercises, they’ll go ‘We want to exercise some conflict. How are we going to move all these thousands of patients?’ Well, what type of patients are they? They’ve all got different requirements and some have different demands on critical care teams. So, if you want to really exercise that, you’ll want to have a capability to be able to generate a variety of patients, and not just do that manually. So the lab is building that out,” DeLapp said. 

This generator capability produces realistic synthetic patient data tied to computer-based scenarios. 

“When there’s a conflict — [with] the pace and the need — you’re going to need tools like that to be able to address those challenges,” he noted.

Up next

In DeLapp’s view, the long-term, flexible nature of the Living Plan is key to enabling the lab and command to access a combination of capabilities that are targeted for — but rapidly evolve with — its complex mission set.

“Because one of the benefits, again, of this partnership and the length of it, is that some of the people on the lab staff have been familiar with [staff like] the Surgeon General — and their processes and their challenges — for several years now, and then also they know where they can help in other areas,” he told DefenseScoop.

Command officials’ next chance to test out the MASCOT suite of decision-support tools being developed and refined by the Massachusetts-based lab will occur later this month, during the Ultimate Caduceus 2025 (UC25) exercise. The event — which is designed to test Transcom’s ability to conduct medical evacuations in both the field and computer-simulated settings — will be hosted in multiple locations this year, including at Travis Air Force Base in California.

“MIT Lincoln Laboratory will generate the patient data based on requirements from the UC25 exercise planners. In the future, this capability will be provided as a tool that planners can use directly,” Anthony said regarding the in-the-works synthetic patient generator.

Another tool in the pipeline for the exercise will streamline patient movement request data reviews at the Transcom Patient Movement Requirements Center.

There’s also a variety of other projects under the Living Plan umbrella that span beyond the command’s medical portfolio.

For one, Lincoln researchers are essentially helping officials in Transcom’s Air Operations Center (AOC) gain capacity to pull key datapoints from massive volumes of instant messages shared between personnel in real-world operations to inform senior leaders.

“A flight manager will chat a message to the crew and the crew will message a chat back. That’s thousands and thousands of chats. So the AOC is working on trying to, one, collect all that, and then two, make sense of all that. And using natural language processing and some generative AI, [they aim to] collect and gain insights into maybe something’s going on in one part of the world — like a runway that is flooded — and it’ll more rapidly inform the senior decision-makers that there’s a problem out there,” DeLapp said. 

“They’re doing that prototyping work and then handing it off to one of the Air Force software factories who then build it out,” he told DefenseScoop.

Due to the unique nature of the initiative, TCAC must evaluate projects to sponsor under the Living Plan each year. DeLapp’s team is now gearing up for their fiscal 2027 review.