Uniformed Services University Launches AI Radiology Training to Strengthen Military Medical Readiness

A first-of-its-kind curriculum at the Uniformed Services University trains military radiology residents to critically evaluate artificial intelligence, mitigating severe staffing shortages and enhancing deployed clinical capabilities. 

Two military medical officers in uniform review and discuss chest X-rays on a computer monitor, simulating the integration of clinical decision support tools in standard radiology workflows.
Military medical personnel review radiological imaging, demonstrating the standard clinical workflows where
the Uniformed Services University aims to integrate artificial intelligence to manage increasing clinical volumes.
(Photo credit: Tom Balfour, USU)

March 26, 2026 by USU External Affairs

To combat critical manning deficits and prepare for the future of military medicine, the Uniformed Services University (USU) and Walter Reed National Military Medical Center have implemented a comprehensive artificial intelligence curriculum for radiology residents. The "Fundamentals of AI in Radiology with Capstone" course spans a range of concepts regarding AI and radiology, from fundamentals and ethics through prompt engineering to practical implementation for administrative, educational, clinical, and research activities. By critically assessing and deploying AI tools, graduates ensure operational force readiness amid a rapidly advancing technological landscape.

Co-directed by Air Force Lt. Col. (Dr.) Justin Peacock and Navy Cmdr. (Dr.) Karl Soderlund, the longitudinal curriculum addresses an urgent operational need. As the commercial sector rapidly integrates algorithmic diagnostic tools, the Department of War requires medical officers capable of navigating this technology without compromising patient care or data security.


How Does AI Mitigate Military Radiology Staffing Shortages?

Currently, the Army and Navy are navigating reduced manning levels in the radiology specialty. By integrating AI into standard workflows, USU aims to manage increasing clinical volumes.

The curriculum specifically targets practical, high-stakes scenarios for deployed medical personnel. If a general radiologist is stationed at a forward operating base without the support of an on-site neuroradiologist, they can utilize AI as a clinical decision support tool for complex neurological pathologies that they may not be familiar with, or potentially as a triage tool for evaluating life-threatening traumatic brain injuries. However, the resident must apply clinical judgment to validate or override the tool's output. Maintaining the human in the loop ensures the highest standard of care and strict accountability.

Air Force Lt. Col. Dr. Justin Peacock, co-director of the artificial intelligence radiology course, delivers a presentation while standing in front of a row of state flags.
Air Force Lt. Col. (Dr.) Justin Peacock, co-director of the AI in radiology curriculum, speaks on the necessity
of dismantling preconceived notions about machine learning by engineering a mindset of healthy skepticism.
(Photo credit: Tom Balfour, USU)

The Microscopic View: Evaluating the Physician

Within the broader course, residents participate in a unique class featuring an agentic AI clinical simulation, developed in collaboration with Dr. Thomas Thesen from the Geisel School of Medicine at Dartmouth University. Learners conduct a consultation for high-risk radioiodine therapy to treat Graves’ disease by interacting directly with an AI programmed to act as the patient. The AI maintains a strict clinical history, understands the simulation parameters, and holds firm behavioral boundaries against user manipulation.

Crucially, the simulation reverses the standard dynamic: the AI evaluates the physician. Utilizing a back-end rubric aligned with Accreditation Council for Graduate Medical Education standards, the system scores the resident on medical knowledge, patient care, professionalism, and patient-centered communication.

"The system is designed to be rigorous in its critique," Peacock said. "Post testing showed it provided accurate critique and improvement suggestions, even for nuclear radiology staff, addressing the common issue of AI being sycophantic.”


Engineering Healthy Skepticism

Formal military portrait of Navy Cmdr. Dr. Karl Soderlund, co-director of the artificial intelligence curriculum for radiology residents at the Uniformed Services University.
Navy Cmdr. (Dr.) Karl Soderlund co-directs
the longitudinal artificial intelligence 
curriculum for radiology residents, addressing
an urgent operational need for the Department
of War. (USU photo)
A primary objective of the syllabus is dismantling preconceived notions about machine learning by engineering a mindset of healthy skepticism. The course directors require residents to abandon the extremes of viewing AI as either entirely useless or as infallible magic.

Increasingly, the Defense Health Agency is investing in AI technologies within the radiology space. Because of this, they require radiology subject matter experts to carefully validate and determine the clinical efficacy of those tools for routine radiologic practice, as well as audit the tools after deployment.

To complete the course, learners must design and present an original capstone project by building an AI-supported tool designed for administration, education, clinical practice, or research. By functioning as developers, residents observe the technology's inherent failure modes firsthand. This hands-on friction prepares future military medical leaders to accurately assess the flaws, biases, and integration challenges of new commercial AI tools proposed for military treatment facilities.

This course prepares radiology faculty, residents, students, technologists, and ancillary support staff for an AI-enabled future in military radiology. It ensures personnel know how to ethically and effectively implement these tools to optimally support the warfighter whenever and wherever they are located.