Physician Cognitive Burden, Telemedicine, and Low-Value Care

Presentation Type: Oral Presentation - Regular
Presentation Time: 03:30 PM - 03:42 PM

Abstract Keywords: Telemedicine, Health at Home, and Virtual Care, Health Policy, Reimbursement and Affordability, and Sustainability, Clinician Well-Being
Primary Track: Advancing Wellness for Providers and Community with Consideration of Human Factors


Using EHR metadata from 449,799 ambulatory encounters, we examined how physician cognitive burden relates to low-value care (LVC) and how telemedicine moderates this relationship. Higher cognitive burden was associated with greater LVC in in-person visits, whereas telemedicine encounters showed the opposite pattern, with increasing burden linked to lower LVC and a clear dose–response effect, suggesting that strategically deploying telemedicine and cognitive burden–aware decision support may help health systems reduce LVC without reducing access.

Speaker(s):
A J Holmgren, PhD
University of California, San Francisco

Author(s):
A J Holmgren, PhD - University of California, San Francisco; Thomas Kannampallil, PhD - Washington University School of Medicine; Julia Adler-Milstein, PhD, FACMI - UCSF School of Medicine;

---

Bridging the Language Gap: Informatics Solutions Using Automation and LLMs to Improve Translated Discharge Instruction Workflows

Presentation Type: Oral Presentation - Regular
Presentation Time: 03:42 PM - 03:54 PM

Abstract Keywords: Outcomes Improvement and Equity, Clinical Decision Support and Care Pathways, Workforce Automation, Communication, and Workflow Efficiency
Primary Track: Implementing Real-World Change, Digital Engagement, and Connected Health


Improving equitable access to translated discharge instructions remains a critical need for families who prefer languages other than English. We implemented workflow automation and supervised LLM-assisted translation to provide timely, consistent Spanish-language materials. By integrating bilingual templates and enabling Language Services to post-edit LLM drafts, translation rates improved from 2 percent to 88 percent without affecting balancing measures, demonstrating a scalable, informatics-driven model for advancing language equity.

Speaker(s):
Osvaldo Mercado, MD
Children's Hospital of Philadelphia

Author(s):
Brooke Luo, MD - Children's Hospital of Philadelphia; Joshua Kurtz, MD - Children's Hospital of Philadelphia; Danielle Capriola, MHA - Children's Hospital of Philadelphia; Stan Krajewski, BBA, MIS - Children's Hospital of Philadelphia; Massiel Ortega Rivera, MHA - Children's Hospital of Philadelphia; Priscilla Ortiz, PhD - Children's Hospital of Philadelphia; Feihan Xin, MSHI - Children's Hospital of Philadelphia;

---

Real-world implementation of AI translation: A Quality Improvement Study in a Bilingual Pediatric Behavioral Health Clinic

Presentation Type: Oral Presentation - Regular
Presentation Time: 03:54 PM - 04:06 PM

Abstract Keywords: Quality Informatics and Lean, Outcomes Improvement and Equity, Workforce Automation, Communication, and Workflow Efficiency, Change Management, Human Factors and Usability, Social Determinants of Health (SDoH)
Primary Track: Implementing Real-World Change, Digital Engagement, and Connected Health


This quality improvement study piloted a novel workflow to safely implement and sustainably monitor a human-in-the-loop AI Spanish translation tool in a bilingual pediatric behavioral health clinic. Over the two months, five providers utilized the tool 27 times with an average user-rated quality of 4.1/5 and one safety concern. Notably, there was a 14% increase in visits in families receiving translated materials, and the majority of families surveyed rated the material as “very useful.”

Speaker(s):
Emily Larimer, MD
University of Washington

Author(s):
Emily Larimer, MD - University of Washington; Jason Lau, MD, MS - University of Washington; Shivika Ahuja, MD - University of Washington; Jacque Burgara, BS - Seattle Children's Research Institute; Arti Desai, MD, MSPH - University of Washington; Michelle Dick, MD - University of Washington; Yu-Hsiang Lin, MD - Seattle Children's Hospital; Lupita Santillan, PhD - University of Washington; K. Casey Lion, MD, MPH - University of Washington;

---

Automated and Human Review of LLM Translated Spanish Discharge Instructions

Presentation Type: Oral Presentation - Regular
Presentation Time: 04:06 PM - 04:18 PM

Abstract Keywords: Workforce Automation, Communication, and Workflow Efficiency, Clinical Decision Support and Care Pathways, Generative AI in Clinical Workflow: Ambient Listening, Chart Summarization, Automated Response with LLM, Outcomes Improvement and Equity, Social Determinants of Health (SDoH)
Primary Track: Implementing Real-World Change, Digital Engagement, and Connected Health


Accurate Spanish-language discharge instructions are crucial for patient safety, yet traditional translation workflows generally require in-person or live video translation, which is resource-intensive and time-consuming, particularly in emergency care settings. Large language models (LLMs) present scalable alternatives, but their translation quality and clinical appropriateness require thorough evaluation. This study investigates LLM-translated Spanish discharge instructions using two evaluation pathways: (a) automated translational metrics and (b) a Multidimensional Quality Metric (MQM)–based LLM-as-judge framework with human validation. We used discharge instructions from the MIMIC dataset to generate Spanish translations with advanced LLMs, then back-translated them into English. In evaluation pathway (a), we computed BLEU, ROUGE, NIST, word error rate (WER), and a concept-based similarity score (CUI) to assess fidelity and surface similarity. For pathway (b), we applied an LLM-as-judge rubric rooted in MQM to score accuracy, fluency, terminology, locale conventions, audience appropriateness, style, and design/markup. A bilingual clinician conducted pass/fail reviews on 20 samples, while a second clinician scored MQM domains to calculate Cohen’s kappa for agreement with LLM-as-judge ratings. Automated metrics favored GPT-o3-mini over Qwen, with higher scores in multiple categories and lower WER. LLM-as-judge MQM scores were consistently high (means 4.87–4.99 on a 6-point scale), with 19 of 20 translations deemed acceptable. Agreement between the clinician and the LLM-as-judge was moderate for locale convention (κ=0.64), and fair for accuracy (κ=0.27) and fluency (κ=0.24), highlighting the need for combined evaluation pathways and ongoing human oversight in multilingual clinical communication.

Speaker(s):
William Mundo, MD, MPH
Department of Emergency Medicine, University of Colorado School of Medicine

Author(s):
William Mundo, MD, MPH - Department of Emergency Medicine, University of Colorado School of Medicine; Maya Kruse, MS - CU Anschutz; Morgan Peña, MPH - CU Anschutz; Elizabeth Goldberg, MD, ScM - Department of Emergency Medicine, University of Colorado School of Medicine; Yanjun Gao, PhD - University of Colorado;

---

Closing Cancer Screening Gaps for Long Stay Psychiatric Inpatients with Serious Mental Illness Through Digital Integration

Presentation Type: Oral Presentation - Regular
Presentation Time: 04:18 PM - 04:30 PM

Abstract Keywords: Innovation Partnerships, Implementation Science, and Learning Health Systems, Clinical Decision Support and Care Pathways, Change Management, Education and Training, Human Factors and Usability
Primary Track: Implementing Real-World Change, Digital Engagement, and Connected Health


A tertiary mental health hospital in Toronto implemented a digital, trauma-informed approach that integrated colorectal, breast, and cervical cancer screening pathways directly into the electronic health record (EHR) to support patient identification, reminders, and follow-up for long-stay inpatients with serious mental illness. A two-year retrospective chart review demonstrated significant improvements in colorectal and cervical screening. These findings suggest that EHR-integrated workflows combined with targeted education can enhance preventive screening in psychiatric inpatient settings.

Speaker(s):
Tania Tajirian, MD
CAMH

Author(s):
Sanjeev Sockalingam; caroline chessex; Cristina de Lasa; Nasrin Adams, MD, PhD - Centre for Addiction and Mental Health; Elnathan Mesfin, MSc; Jennifer Nicolle, CAMH - CAMH;

---

The Power of Pet-Patient Registries to Explore One Health Questions

Presentation Type: Oral Presentation - Regular
Presentation Time: 04:30 PM - 04:42 PM

Abstract Keywords: Analytics, Registries, and the Digital Command Center, Health Data Science, Public Surveillance and Reporting, Environmental Exposure, & Global Health
Primary Track: Big Data for Health


Data from pets receiving veterinary care represents an underutilized resource for accelerating translational science and precision medicine. The Companion Care Registry, a standardized, HIPAA-compliant database linking human and veterinary medical records at the household level, is a groundbreaking approach and model for overcoming translational research challenges. By focusing on data standardization, privacy, and governance, the registry enables advanced cross-species analytical capabilities to investigate shared environmental factors, disease pathways, and novel interventions.

Speaker(s):
Tracy Webb, DVM, PhD
Colorado State University

Author(s):
Tracy Webb, DVM, PhD - Colorado State University; Adam Kiehl, Master of Applied Statistics - Colorado State University; Kathleen Mullen, Postdoctoral fellow/DVM, MS - CU Anschutz; Nadia Saklou, DVM, PhD - Colorado State University; Ian Brooks, PhD - University of Maryland Institute for Health Computing; Sue VandeWoude, DVM - Colorado State University; Joshua Stern, DVM, PhD - North Carolina State University; Andrey Soares, PhD - University of Colorado School of Medicine; Melissa Haendel, PhD - University of North Carolina at Chapel Hill;

---