Biomedical Informatics Commons^™

Evaluation Of Large Language Models As Decision Support Tools For Head And Neck Cancer Management: A Blinded Multidisciplinary Simulation Study, Sholem Hack, Ron J. Karni, Antonino Maniaci, Christopher E. Fundakowski, Luca Castellani, Fabiola Incandela, Remo Accorona, Miguel Mayo-Yanez, Martina Violati, Lorenzo Giannini, Niccolo' Mevio, Alberto Maria Saibene

Department of Otolaryngology - Head and Neck Surgery Faculty Papers

BACKGROUND: The management of head and neck cancer relies on multidisciplinary expertise; however, access to tumor boards remains variable. Large language models (LLMs) may support guideline-based decision-making, although performance in complex oncologic scenarios is not well defined.

METHODS: Fourteen synthetic cases based on real tumor board encounters were evaluated. Five blinded comparator arms produced recommendations: a human expert, Non-RAG-GPT-4, Non-RAG-GPT-5, RAG-GPT-4, and RAG-GPT-5. Eight head and neck oncologic surgeons scored each recommendation for appropriateness, clarity, specificity, and feasibility using 5-point Likert scales. Paired permutation testing and inter-rater reliability were assessed.

RESULTS: LLM outputs showed close alignment with expert recommendations. RAG-based …

Leveraging Informatics To Manage Lifelong Monitoring In Childhood Cancer Survivors, Kimberly Davidow, Renee Gresh, E. Anders Kolb, Ellen Guarnieri, Mary R. Cooper

College of Population Health Faculty Papers

Background: Electronic health records (EHR) have long held promise for sharing information efficiently, but this remains challenging. This quality improvement initiative sought to improve the accurate documentation of anthracycline and radiation therapy exposures in pediatric oncology patients who were treated at different institutions through a quality improvement methodology and EHR tools. Methods: A custom-built EHR smartform was previously created. Modifications were made to the smartform, and quality improvement methods were utilized to improve receipt of radiation summaries from other institutions and documentation of chemotherapeutic doses. Results: Three months after interventions, including clinician education and smartform updates, accurate anthracycline documentation improved …

Influences On Emergency Clinician Use Of Health Information Exchange: Interview Study, Brian E. Dixon, Umesh Ghimire, Benjamin Richter, Corinne Bowditch, Saurabh Rahurkar, John T. Finnell, Joshua R. Vest

Division of Internal Medicine Faculty Papers & Presentations

BACKGROUND: Health information exchange (HIE) supports clinical decision-making in emergency medicine settings. Despite evidence and policies that encourage the adoption of HIE, use by clinicians is limited. Moreover, few studies examine HIE use years after adoption by hospitals or clinics.

OBJECTIVE: This study aims to examine the perceptions and use of a mature, operational HIE system by emergency department clinicians years after its implementation.

METHODS: We interviewed 21 clinicians in various roles (eg, attending physician and nurse practitioner) across multiple health systems that participate in a statewide HIE network. We asked questions about their use of the HIE system and …

Large Language Models As Information Providers For Appropriate Antimicrobial Use: Computational Text Analysis And Expert-Rated Comparison Of Chatgpt, Claude And Gemini, Marcello Di Pumpo, Maria Rosaria Gualano, Danilo Buonsenso, Francesca Raffaelli, Daniele Donà, Vittorio Maio, Patrizia Laurenti, Walter Ricciardi, Leonardo Villani

College of Population Health Faculty Papers

OBJECTIVES: Antimicrobial resistance is a critical public health threat. Large language models (LLMs) show great capability for providing health information. This study evaluates the effectiveness of LLMs in providing information on antibiotic use and infection management.

METHODS: Using a mixed-method approach, responses to healthcare expert-designed scenarios from ChatGPT 3.5, ChatGPT 4.0, Claude 2.0 and Gemini 1.0, in both Italian and English, were analysed. Computational text analysis assessed readability, lexical diversity and sentiment, while content quality was assessed by three experts via DISCERN tool.

RESULTS: 16 scenarios were developed. A total of 101 outputs and 5454 Likert-scale (1-5) scores were obtained …

Intelligence Architectures And Machine Learning Applications In Contemporary Spine Care, Rahul Kumar, Conor Dougherty, Kyle Sporn, Akshay Khanna, Puja Ravi, Pranay Prabhakar, Nasif Zaman

SKMC Student Presentations and Publications

The rapid evolution of artificial intelligence (AI) and machine learning (ML) technologies has initiated a paradigm shift in contemporary spine care. This narrative review synthesizes advances across imaging-based diagnostics, surgical planning, genomic risk stratification, and post-operative outcome prediction. We critically assess high-performing AI tools, such as convolutional neural networks for vertebral fracture detection, robotic guidance platforms like Mazor X and ExcelsiusGPS, and deep learning-based morphometric analysis systems. In parallel, we examine the emergence of ambient clinical intelligence and precision pharmacogenomics as enablers of personalized spine care. Notably, genome-wide association studies (GWAS) and polygenic risk scores are enabling a shift from …

Refine: A Database Of Linked Clinical Data And Genomic Biomarkers In Renal Cell Carcinoma Patients Receiving Immunotherapy-Based Treatment Regimens, Jeffrey Zhong, Albert Jang, Bashar Abuqayas, Arnab Basu, David Benjamin, Vineel Bhatlapenumarthi, Mehmet Asim Bilen, Dhvani Buch, Mark Chang, Erica Chin, Sourat Darabi, Nagendra Dhanikonda, Pooja Ghatalia, Claud Grigg, Abby Grier, Tanya Jindal, Joannah Jung, Deepak Kilari, Hamsa Kumar, Suzanna Lee, Brittany Neelands, Chinmayi Pandya, Jeff Pawalek, Jaimee Staggers, Ahmet Yildirim, Yousef Zakharia, Kevin Zarrabi, Michael Zimmerman, George Sledge, David Spetzler, Andrew Elliott, Rana Mckay, Pedro Barata

Department of Medical Oncology Faculty Papers

The REnal cancer consortium for Focused Investigation of Novel biomarkers and Expression (REFINE) consortium represents an important initiative in integrating clinical data with molecular sequencing in patients with advanced renal cell carcinoma (RCC) treated with immunotherapy-based approaches. By leveraging real-world evidence and genomic analysis, this consortium aims to explore putative predictive biomarkers with the potential to inform personalized treatment strategies. Findings from the REFINE database may further contribute to our understanding of disease courses of immunotherapy-based approaches for various molecular subtypes of RCC, associations of race and ethnicity with RCC treatment and outcomes with representation of patient populations underrepresented in …

Deep Learning-Based Auto-Segmentation For Liver Yttrium-90 Selective Internal Radiation Therapy, Jun Li, Wookjin Choi, Rani Anne

Department of Radiation Oncology Faculty Papers

The aim was to evaluate a deep learning-based auto-segmentation method for liver delineation in Y-90 selective internal radiation therapy (SIRT). A deep learning (DL)-based liver segmentation model using the U-Net3D architecture was built. Auto-segmentation of the liver was tested in CT images of SIRT patients. DL auto-segmented liver contours were evaluated against physician manually-delineated contours. Dice similarity coefficient (DSC) and mean distance to agreement (MDA) were calculated. The DL-model-generated contours were compared with the contours generated using an Atlas-based method. Ratio of volume (RV, the ratio of DL-model auto-segmented liver volume to manually-delineated liver volume), and ratio of activity (RA, …

Moud 2.0: A Clinical Algorithm And Implementation Evaluation Protocol For Sublingual And Injectable Buprenorphine Treatment Of Opioid Use Disorder, Brandon Joa, Eric Fung, Michael Weinstein, Lara Weinstein

Department of Family & Community Medicine Faculty Papers

BACKGROUND: Primary care is the initial contact point for most patients with opioid use disorder (OUD) but lacks tools for guiding treatment. Only a small fraction of patients access evidence-based care. Long-acting injectable buprenorphine has potential to improve medication adherence and program retention in low-barrier primary care treatment settings. We present the first clinical decision support algorithm incorporating long-acting buprenorphine (LAIB) in primary care. We include a protocol for a future evaluation of the algorithm's implementation process, "Medication for Opioid Use Disorder (MOUD) 2.0," at a housing and integrated care clinic at a Federally Qualified Health Center.

METHODS: Literature review …

Impact Of An Online Decision Support Tool For Ductal Carcinoma In Situ (Dcis) Using A Pre-Post Design (Aft-25), Elissa Ozanne, Kellyn Maves, Angela Tramontano, Thomas Lynch, Alastair Thompson, Ann Partridge, Elizabeth Frank, Deborah Collyar, Desiree Basila, Donna Pinto, Terry Hyslop, Marc Ryser, Shoshana Rosenberg, E. Shelley Hwang, Rinaa Punglia

Department of Pharmacology and Experimental Therapeutics Faculty Papers

BACKGROUND: The heterogeneous biology of ductal carcinoma in situ (DCIS), as well as the variable outcomes, in the setting of numerous treatment options have led to prognostic uncertainty. Consequently, making treatment decisions is challenging and necessitates involved communication between patient and provider about the risks and benefits. We developed and investigated an interactive decision support tool (DST) designed to improve communication of treatment options and related long-term risks for individuals diagnosed with DCIS.

FINDINGS: The DST was developed for use by individuals aged > 40 years with DCIS and is based on a disease simulation model that integrates empirical data and …

Changing Landscape Of Liver Transplant In The United States—Time For A New Innovative Way To Define And Utilize The “Non-Standard Liver Allograft”—A Proposal, Rashmi Seth, Kenneth Andreoni

Department of Surgery Faculty Papers

Since the first liver transplant was performed over six decades ago, the landscape of liver transplantation in the US has seen dramatic evolution. Numerous advancements in perioperative and operative techniques have resulted in major improvements in graft and patient survival rates. Despite the increase in transplants performed over the years, the waitlist mortality rate continues to remain high. The obesity epidemic and the resultant metabolic sequelae continue to result in more marginal donors and challenging recipients. In this review, we aim to highlight the changing characteristics of liver transplant recipients and liver allograft donors. We focus on issues relevant in …

The Design Of A Quality Improvement Dashboard For Monitoring Spinal Cord And Column Injuries, Zahra Azadmanjir, Mohsen Sadeghi-Naini, Mohammad Dashtkoohi, Maziar Moradi-Lakeh, Jalil Arabkheradmand, James Harrop, Vafa Rahimi-Movaghar

Department of Neurosurgery Faculty Papers

Background: Interactive dashboards are a powerful tool for dynamic visualization and monitoring of patient performance and serve as a useful to for optimal decision-making. The National Spinal Column and Cord Injury Registry of Iran (NSCIR-IR) was designed to efficiently display and broadcast important patient care data. This has been achieved through an electronic dashboard display (graph and visual displays), rather than traditional static paper reports (text). Objectives: The objective of this study was to design and develop an electronic visual dashboard as a display system to monitor the quality of care in the NSCIR-IR collaborating centers. Methods: The indicators chosen …

Using Metabolic Potential Within The Airway Microbiome As Predictors Of Clinical State In Persons With Cystic Fibrosis, Gabriella Shumyatsky, Aszia Burrell, Hollis Chaney, Iman Sami, Anastassios C Koumbourlis, Robert J Freishtat, Keith A Crandall, Edith T Zemanick, Andrea Hahn

Department of Medical Laboratory Sciences & Biotechnology Faculty Papers

Introduction: Pulmonary exacerbations (PEx) in persons with cystic fibrosis (CF) are primarily related to acute or chronic inflammation associated with bacterial lung infections, which may be caused by several bacteria that activate similar bacterial genes and produce similar by-products. The goal of our study was to perform a stratified functional analysis of bacterial genes at three distinct time points in the treatment of a PEx in order to determine the role that specific airway microbiome community members may play within each clinical state (i.e., PEx, end of antibiotic treatment, and follow-up). Our secondary goal was to compare the change between …