George Mason University’s Injury Analytics Lab announced on August 12, 2026 that it has been awarded a patent for an AI-driven system designed to detect and assess bruises, burns, scars, birthmarks, rashes and other skin conditions with far more consistency across different skin tones than clinicians have historically been able to achieve with the naked eye. The technology combines computer vision with electronic health record data, and its developers say the goal is as much about equity in medical and legal evidence as it is about diagnostic accuracy.
A Documentation Problem With Real Consequences
Bruises are one of the most common pieces of physical evidence in cases of interpersonal violence, child abuse and elder abuse, and they are also one of the hardest injuries to document reliably. Lighting conditions, the age of the bruise, and above all a patient’s skin tone all affect how visible a bruise is to the human eye, and research in forensic nursing has repeatedly found that bruises on darker skin are detected less often and rated as less severe than clinically equivalent injuries on lighter skin, a gap with direct consequences for both medical treatment and legal cases that hinge on injury documentation. Katherine Scafide, an associate professor of nursing and co-director of the Injury Analytics Lab, framed the stakes plainly, saying that documenting injuries accurately is essential for both medical treatment and seeking justice.
How the System Works
The patented technology automatically adjusts light wavelengths, camera filters and imaging settings based on an individual’s skin tone before capturing an image, rather than applying the same fixed imaging protocol to every patient regardless of complexion. It then layers that image data with information pulled from the patient’s electronic health record, including clinical history, medications, demographics and existing skin characteristics, to produce an assessment of a bruise’s age, size, color and shape along with a confidence level for each estimate. The team has compared the concept to plant identification apps that combine an image with location data to narrow down a species, applying the same logic of pairing visual data with contextual information to sharpen an assessment that a photo alone could not reliably produce.
Who Built It
The patent is the product of a multi-year, cross-disciplinary collaboration inside George Mason’s Injury Analytics Lab, co-directed by Scafide alongside Janusz Wojtusiak, a professor of health informatics, and David Lattanzi, a professor of engineering. That combination of nursing, health informatics and engineering expertise reflects how much of the technical difficulty in this project sits outside pure image recognition, in reconciling clinical judgment, imaging physics and machine learning into a single practical tool. The lab has built a repository of more than 100,000 images spanning bruises, burns, scars, birthmarks, rashes, eczema and psoriasis cases, the training and validation base behind the patented system.
Part of a Longer Research Arc
This patent builds on work the lab has been doing publicly for at least two years. In 2025, the same group launched a mobile app that lets survivors of interpersonal violence capture and document their own bruises, a tool aimed at giving people evidence they control directly rather than relying solely on a clinical encounter that may happen days after an injury when a bruise has already faded or changed color. The lab has also drawn federal backing for this line of work, having previously secured funding through the NIH’s AIM-AHEAD program, an initiative specifically focused on improving equity in AI-driven health research, underscoring that the disparities this system targets have already drawn attention and funding at the national level.
Where the Skepticism Sits
Tools that combine forensic documentation with AI inference carry an obvious tension: the same system built to strengthen evidence for survivors of violence could, if its confidence estimates are miscalibrated, produce assessments that are challenged or dismissed in legal proceedings where the underlying algorithm and its training data face scrutiny from opposing counsel. Clinicians and legal advocates working in this space have long emphasized that any tool used to support abuse investigations needs to be transparent about its uncertainty rather than presenting a single definitive answer, particularly given how much is riding on injury assessments in custody disputes, criminal cases and protective order hearings. The Injury Analytics Lab’s decision to report confidence levels alongside each estimate, rather than a flat yes-or-no determination, suggests the team is aware of that scrutiny, though how courts and clinical review boards will treat those confidence scores in practice remains untested at scale.
What Comes Next
Having secured the patent, the George Mason team says the next step is bringing the technology into actual healthcare, forensic and research settings, moving it from a research prototype into tools clinicians, forensic nurses and investigators can use on real cases. If the system performs as designed, it could narrow a documentation gap that has disadvantaged patients with darker skin tones for as long as bruise assessment has relied on visual inspection alone, giving both emergency departments and courts a more consistent, standardized basis for judging injuries that have historically been assessed differently depending on who was looking and what their equipment could see.
