Nursing Research

Can AI help make medical records less biased? New study suggests yes—with caveats

Heather Carroll — Fri, 10 Jul 2026 16:50:49 +0000

Can AI help make medical records less biased? New study suggests yes—with caveats Heather Carroll Fri, 07/10/2026 - 12:50

Large language models can identify judgmental language in clinical notes but the settings play a major role in accuracy.

Teenu Xavier PhD, RN. Photo provided

“Addict,” “non-compliant,” “failed treatment,” and “obese person” are examples of stigmatizing language that can appear in medical records. At 91��’s College of Public Health, researchers are exploring whether artificial intelligence (AI) can help identify this kind of language in clinical notes before it impacts patient care.

Nurse scientist Teenu Xavier and colleagues found that large language models (LLMs) show promise in identifying stigmatizing language in clinical documentation, but their performance is highly dependent on their settings. Model size, temperature settings, prompting strategies, and even note type can substantially influence results.

One finding was consistent across every model tested: Providing examples of stigmatizing language improved accuracy.

“Simply selecting an LLM is not enough when used for clinical documentation,” said Xavier, an assistant professor in the School of Nursing. “Careful attention must be paid to settings and prompting before these tools can be reliably used in health care environments.”

Why does this matter?

The use of stigmatizing language in clinical documentation can reinforce bias and affect a patient’s future care. AI tools may be able to help identify this kind of language, promoting more equitable care and improving patient trust and experience.

“Pre-trained models, when optimized for identifying stigmatizing language, could help enable more timely interventions and modifications to the documentation process,” Xavier said. “Our research highlights the need for continued collaboration between health care professionals and AI developers to create tools that improve communication, reduce bias, and improve the overall patient experience.”

What are the detailed study findings?

The largest LLM (trained on large amounts of data) was the best at predicting “stigmatizing” language (94%), but the worst at predicting “not stigmatizing” correctly (47%).
The smallest LLM was the best at predicting “not stigmatizing” correctly (99.7%), but worst at predicting “stigmatizing” correctly (2%).
When researchers gave the LLM an example of stigmatizing language, accuracy improved in all models.
Emergency provider notes were most accurately (69%) categorized as “stigmatizing” or “non-stigmatizing,” and plan of care notes had the lowest accuracy (56%). Misclassifications most commonly arose in long, clinically dense notes where neutral descriptions of complex illness or adverse events were mistaken by the models for judgmental language.
Larger models worked best at lower temperature (how predictable or random the LLMs’s output is when making a classification) and smaller models improved with higher temperature, which means the LLM took more risks in interpretation.

“Detecting stigmatizing language with large language models: mind the settings” was published in JAMIA Open in April 2026. Co-authors include Jane M. Carrington from the University of Florida and Joshua Lambert from the University of Cincinnati.

Thumbnail photo by issaronow from Adobe Stock.

Key Takeaways

A George Mason study found that large language models show promise in identifying stigmatizing language in clinical documentation, but their performance is highly dependent on their settings, such as model size, temperature settings, prompting strategies, and note type.
The ability to detect and correct stigmatizing language early can reduce bias in patient care and lead to improved patient trust and better health outcomes.
Continued collaboration between health care workers and AI developers is needed to create tools that enhance communication, reduce bias, and improve the overall patient experience.

Learn about Artificial Intelligence at George Mason

Topics

large language models

Machine Learning in Health Care

Health Care Services

College of Public Health

Artificial Intelligence

Research

CPH research

Nursing Research

Sleep problems in dementia care linked more to emotional well-being than fear of falling

Heather Carroll — Tue, 07 Jul 2026 17:27:48 +0000

Sleep problems in dementia care linked more to emotional well-being than fear of falling Heather Carroll Tue, 07/07/2026 - 13:27

Key Takeaways

For dementia care recipients, positive emotion was associated with better sleep, while psychological distress, like depression and anxiety, was associated with worse sleep among caregivers.
Though the study found that fear of falling was associated with poorer sleep for both partners, emotional well-being ultimately showed a stronger association with sleep than fear of falling itself.
This study by 91�� researcher Kyeung Mi Oh is the first to examine the relationships between fear of falling in older adults with probable or possible dementia and the subjective sleep quality of both care recipients and their spousal caregivers.

In This Story

Kyeung Mi Oh, PhD, MSN, RN, FAAN

While fear of falling has long been linked to poor sleep among older adults, new research from 91�� suggests emotional well-being may be an even stronger predictor of sleep quality for people with dementia and the spouses who care for them.

Dementia researcher Kyeung Mi Oh, professor in the School of Nursing. Photo provided.

In what is believed to be the first study of its kind, dementia researcher Kyeung Mi Oh examined how fear of falling and sleep quality interact within dementia caregiving couples where one partner has probable or possible dementia and the other serves as caregiver.

Falls are a common concern among older adults and can lead to reduced mobility, more frequent hospital visits, and limits on daily activities. Many individuals also develop a persistent fear of falling, causing them to avoid activities even when they are physically able. In dementia care, that fear can further increase caregiver burden and affect sleep for both partners.

Though the study found that fear of falling was associated with poorer sleep for both partners, emotional well-being ultimately was more strongly linked to sleep outcomes.

Why does this matter?

“These findings suggest that improving emotional well-being may be more important for sleep health in dementia care dyads than addressing physical health factors alone,” said Oh, professor of Nursing in 91��’s College of Public Health.

For care recipients, positive emotion was associated with better sleep, while psychological distress, like depression and anxiety, was associated with worse sleep among caregivers.

The results highlight the importance of assessing both physical and emotional factors in dementia care.

“Routine assessment of fear of falling, fall history, and psychological distress in both members of dementia care couples could help identify those at risk for poor sleep,” said Oh. “Clinically, interventions that promote emotional well-being and reduce psychological distress may improve sleep for both care recipients and caregivers, highlighting the need for a family-centered approach to dementia care.”

Key Findings:

Fear of falling was associated with poor sleep; however, after factoring in demographic, lifestyle, and health-related factors, positive emotion predicted lower insomnia severity in care recipients, while psychological distress predicted higher insomnia severity in caregivers.
Both fear of falling and recent falls were linked to worse insomnia for people with dementia and their caregivers.

Yeonsu Song from the University of California, Los Angeles co-authored the paper, “Fear of Falling and Sleep Disturbances in Spousal Care Dyads Affected by Dementia,” which was first published online in January 2026 in the Journal of Applied Gerontology.

Thumbnail photo By Koto Amatsukami from Adobe Stock.

Topics

College of Public Health

CPH research

Nursing Research

Groundbreaking mobile app captures and documents bruises to help survivors of interpersonal violence

mthomp7 — Thu, 05 Jun 2025 16:47:09 +0000

Groundbreaking mobile app captures and documents bruises to help survivors of interpersonal violence mthomp7 Thu, 06/05/2025 - 12:47

Anonymous donor commits additional $5.3 million to advance research

An interdisciplinary 91�� research team is breaking new ground in using artificial intelligence to support victims of interpersonal violence. Led by Kat Scafide and Janusz Wojtusiak of the College of Public Health and David Lattanzi of the College of Engineering and Computing, the Evidence-based AI Software for Injury Detection (EAS-ID) project has successfully completed Phase 1: development of a working prototype of a mobile app designed to accurately capture and document bruises.

The tool has the potential to transform how clinicians and frontline professionals identify, record, and communicate evidence of injury, particularly in cases of interpersonal violence.

The EAS-ID (Evidence-based AI Software for Injury Detection) project has successfully completed Phase 1: development of a working prototype of a mobile app designed to accurately capture and document bruises. An anonymous donor recently committed an additional $5.3 million to advance the technology. Photo by Mary Cunningham/College of Public Health

The EAS-ID app acts as a digital guidance system—similar to mobile check-deposit apps that help users capture clear, usable images. When a clinician uses the tablet-based tool, the app detects the presence of a bruise and provides real-time guidance to ensure the image meets clinical and legal standards. A rectangle tracks the injury area, much like facial recognition technology, ensuring the final image captures exactly what’s needed to document the bruise effectively.

The research team announced $4.85 million in initial funding from the same anonymous donor in March 2024.

“Initially, we were laser-focused on detection—how to identify whether a bruise was present,” said Scafide, a forensic nurse. “But Phase 1 taught us that detection is only a small part of the clinical and legal challenge. The real complexity lies in documentation.”

That documentation must be both clinically sound and legally reliable. Forensic nurses—the app’s initial target users—often assess injuries while juggling multiple tasks: shining a light, holding a camera, completing paperwork. The EAS-ID tool reduces that burden by enabling simultaneous photo capture and structured documentation.

The app’s workflow is tailored to the real-world conditions of forensic nursing. It walks users through a standardized documentation process built directly into the interface, making it possible to create accurate, consistent records that can hold up in court—even years later.

“The documentation that nurses create is reviewed by many others in the criminal justice system,” noted Lattanzi, a civil engineer. “It must be reliable—not a liability.”

To ensure the app meets forensic standards, the team conducted extensive focus groups with practicing forensic nurses. These sessions were crucial to understanding not just what needed to be documented, but how clinicians make decisions in real time under pressure. Their insights helped shape the app into a useful tool with intuitive guide rails and legal relevance.

Based on Phase 1 progress, the team is preparing to broaden its focus to include a wider range of users. “We’re exploring how to adapt the app for clinicians—and even non-clinicians—who don’t have a forensic background but still need to document injuries,” said Scafide.

Wojtusiak, a machine learning expert, added: “Think of it like TurboTax. You don’t need to be a CPA to file your taxes. We want to make it possible for someone who sees fewer IPV cases—or who isn’t an expert in injury documentation—to still collect reliable evidence.”

Behind the app, a powerful AI engine is being trained to distinguish bruises from other skin discolorations and guide users through appropriate documentation steps. To train the AI, the team is building a massive, diverse dataset. So far, they’ve partnered with Inova Health System and Adventist Healthcare Shady Grove Medical Center to collect tens of thousands of images across different skin tones, body types, and medical conditions.

“Deep learning requires scale—and the team has set a goal of collecting one million images,” said Wojtusiak.

This summer, the team will launch a crowdsourcing initiative to expand the dataset beyond clinical settings. “All of our current data was collected in labs, under controlled conditions,” explained Scafide. “But real-world data is messier. We need to understand what the photos and data will look like when collected by the public, outside of protocols.”

The app’s architecture was designed with that in mind. Using machine learning, the app evolves over time—each new image improves its performance. This allows for continual refinement, even after the app is deployed.

The project has moved at an unusually fast pace for an academic endeavor, thanks in part to the support of students and research assistants whose contributions have accelerated development while gaining invaluable experience. Continued progress has been made possible by a second round of funding from the same anonymous donor.

With a patent pending on the app’s innovative integration of image capture and health assessment data, the EAS-ID team is now preparing to scale development, deepen partnerships, and explore commercialization pathways.

“Our goal is to get this technology into the hands of people who need it—quickly,” said Scafide. “The faster we can deploy, the faster we can support the professionals working to protect and care for victims.”

Grand Challenge Initiative

Explore George Mason’s Grand Challenge Initiative addressing the greatest threats to humanity’s ability to live a peaceful, healthy, prosperous, and just existence.

“Think of it like TurboTax. You don’t need to be a CPA to file your taxes. We want to make it possible for someone who sees fewer IPV cases—or who isn’t an expert in injury documentation—to still collect reliable evidence.”

In This Story

Katherine Scafide, PhD, RN, FAAN, FAAFS

Janusz Wojtusiak, PhD

David Lattanzi

Topics

EAS-ID

Bruise Visibility

Research

College of Public Health

College of Engineering and Computing

CEIE Success Stories

GCI-Grand Challenge Initiative

Student nurses converge for an interdisciplinary research conference organized by George Mason students

Taylor Thomas — Wed, 04 Jun 2025 13:59:11 +0000

Student nurses converge for an interdisciplinary research conference organized by George Mason students Taylor Thomas Wed, 06/04/2025 - 09:59

In This Story

Karen Trister Grace, PhD, MSN, CNM, FACNM

Tara Altay, a PhD student in George Mason’s School of Nursing, noticed an unfortunate conference trend: a decrease in attendance and an increase in price. Disheartened but refusing to remain passive, Altay leveraged her role as treasurer of the 91�� Graduate Nursing Organization (GNO) to combat this observation becoming the norm. Refusing to see knowledge and network become inaccessible to fellow students and nursing professionals, the GNO, in partnership with George Mason Nurses for Sexual and Reproductive Health (NSRH), organized a D.C.-area interdisciplinary student nursing research conference.

Tara Altay. Photo provided

“The main motivation was to create an affordable, in this case free, student research conference inclusive of all nursing degree programs (ADN, BSN, MSN, DNP, PhD),” Altay said. "One of the most fun parts of professional conferences is the connections made and side conversations with peers. We wanted to create that opportunity for students here and all students around D.C., Maryland, and Virginia."

Altay also added that they also wanted to provide doctoral students with a supportive space to practice presenting their research—one where peers fostered meaningful discussions that enriched and strengthened their work.

The 67 student participants who registered for the conference came from the University of Maryland, Shenandoah University, University of Virginia, George Washington University, Johns Hopkins University, and Virginia Commonwealth University.

“This conference was a great networking opportunity, as well as a chance to learn what other doctoral students are doing during their final year project and what they plan to do with their degree afterwards,” said Kathryn Oliver, an attendee and Doctor of Nursing Practice (DNP) Family Nurse Practitioner (FNP) student.

As a self-described young, aspiring health care professional of color, Ka'Lise Benson, a direct-entry Bachelor of Science in Nursing (BSN) student who is a University Scholar in the Honors College, was thrilled to engage with those with similar aspirations. “I left feeling so grateful and so encouraged. I felt that the conference, having placed me in a space full of like-minded individuals, confirmed the call I already felt so led to answer,” she said.

Meghan Stough presenting her research on opioid use disorder interventions. Photo by Rene Ayala

The conference featured doctoral research in both poster and podium presentations, offering bachelor’s and master's-level students the opportunity to observe the trajectory of their future careers. Meghan Stough, another DNP-FNP student, was one of 18 student presenters. She utilized the conference to share her DNP project on evidence-based interventions to expand access to medications for opioid use disorder in the pre-hospital setting.

“The conference allowed me to practice translating complex clinical work into a concise, accessible format—an essential skill in nursing leadership and advocacy. Presenting and receiving feedback enhanced my confidence in scholarly communication and solidified my role as a nurse leader committed to knowledge dissemination,” Stough said.

Community practitioners were invited to share their field experiences and the innovative ways they apply their nursing degrees to serve diverse populations beyond hospital and primary care settings, including keynote speaker Terri-Ann Thompson from Ibis Reproductive Health, who spoke about her work on reproductive justice. Thompson was joined by four nurse practitioners as co-panelists in breakout sessions. Additionally, the College of Humanities and Social Sciences Women and Gender Studies department adapted their Intersectionality 101 program into a nurse-specific presentation and discussion panel for attendees in support of the conference’s theme, “Equity.”

“I'm so proud of our student chapter members for organizing this impressive event. It was inspiring to spend the day among such motivated students, and I hope this is just the start of an ongoing tradition at George Mason,” said Karen Trister-Grace, assistant professor in the School of Nursing and faculty advisor for NSRH.

Attendees from across the Washington, D.C.-area gather for George Mason student-led nursing conference. Photo by Rene Ayala

Topics

Nursing Research

Student Research

Master of Science in Nursing

bachelor of science in nursing

Doctor of Nursing Practice (DNP)

Students

Nurses trained via virtual reality performed better than those trained by inpatient clinical training

Mary Cunningham — Tue, 11 Apr 2023 20:37:23 +0000

Nurses trained via virtual reality performed better than those trained by inpatient clinical training Mary Cunningham Tue, 04/11/2023 - 16:37

Lab Coordinator Bethany Cieslowski helps nursing student Katherine Gugino adjust the virtual reality headset during a VR Simulation Lab training class. Photo by: Shelby Burgess, Office of University Branding

In search of novel tactics to accommodate a larger student body and fulfill workforce demands, nursing schools are developing new approaches to optimize learning, engage learners, and provide methods to ensure competency in future nursing graduates.

A recent study by Bethany Cieslowski, associate professor of Nursing, and colleagues found that immersive virtual reality (VR) training has been shown to be as effective as inpatient training for students learning to provide care for acute care pediatric patients. The performance of student nurses who underwent immersive virtual reality training was found to be better overall than that of their counterparts who received inpatient clinical training.

“The use of VR continues to grow in clinical nursing practice; however, the existing body of evidence on VR, especially immersive virtual reality, is limited. These preliminary results offer promise and demonstrate the potential of immersive virtual reality in the future of nursing education and preparing the workforce of the future,” said Cieslowski.

Nursing student Elianna Puerto uses a virtual reality set for a new simulation platform, SimX, during a VR Simulation Lab training class. Photo by: Shelby Burgess, Office of University Branding

The average (mean) scores for students in the immersive VR training group were higher for all performance domains, with particularly notable improvements in the subdomains of infection control, initial assessment, and oxygen therapy compared to the clinic trained group. There were no significant differences between the groups in focused assessment, medication administration, and evaluation. Total performance scores were significantly higher for the VR group.

The mixed-methods without randomization pilot study examined the effectiveness of an immersive VR simulation training program to replace pediatric clinical practice for a convenience sample of junior-level prelicensure nursing students.

“The Development and Pilot Testing of Immersive Virtual Reality Simulation Training for Prelicensure Nursing Students: A Quasi-Experimental Study” was published in Clinical Simulation in Nursing in February 2023. Tanya Haas, assistant professor of nursing; Kyeung Mi Oh, associate professor of nursing; Kathleen Chang, assistant professor of Nursing; and Cheryl Oetjen, interim director of the School of Nursing were also part of the research team.

The Medical Team, Inc. Endowment for Innovations in Nursing Education Fund funded part of the research.

Immerse yourself in a Nursing Program at Mason

Apply to Mason