CDS for Pneumonia Diagnosis
Michael Ward is leading an AHRQ-funded grant to implement cutting edge AI-based image processing technology and clinical decision support in the VUMC ED to rapidly diagnose pneumonia and ensure prompt evidence-based treatment.
LLMs for Identifying Emergent Patient Portal Messages
Siru Liu recently published a paper demonstrating how a knowledge graph can enhance large language models' ability to identify patients with emergent conditions on patient portals. Currently, these portals display only a general disclaimer advising patients to seek emergency care when necessary. However, she has observed messages (e.g., "an elephant sitting on my chest") that clearly indicate an emergency condition. Her new tool could refine the generic disclaimer into a personalized emergency notification based on the content of the patient message, potentially making a significant impact on life-threatening situations.
Ophthalmology Informatics
Eric Brown has been leading and contributing to a wide range of clinical informatics projects at the intersection of ophthalmology and data science, including standardizing visual acuity concepts in SNOMED for the OMOP CDM to allow collaborations with different EHRs to use eyecare clinic data, developing machine learning tools for thyroid eye disease detection, and generating synthetic ophthalmic imaging datasets for AI research. Additional initiatives include automating data extraction for retinopathy of prematurity, enhancing structured documentation in eStar for OCT imaging, analyzing GLP-1A safety signals, and supporting award-winning resident and fellow research through data extraction and statistical collaboration.
CDS for Genomics
Josh Peterson has contributed to genomics-focused clinical decision support efforts, including the GUARDD project, which delivers tailored recommendations for APOL1-positive adults at risk for kidney disease. He also supported an innovative initiative that identifies actionable findings on echocardiograms and routes those patients for appropriate genetic evaluation, helping to close gaps in care and integrate precision medicine into routine clinical workflows.
Improved Patient Education on Test Results
Bryan Steitz has led work to improve how patients access and interpret their test results through My Health at Vanderbilt. Since VUMC began making all results immediately available in 2021, his work has focused on developing and evaluating strategies that support patient understanding and reduce anxiety. These initiatives include changes to default notification preferences and educational materials presented alongside test results—efforts that have helped over 90% of patients engage directly with their health information in real time.
Brim Analytics
Dan Fabbri developed and launched Brim Analytics, an AI-enhanced tool to extract structured data from unstructured medical records. This product has been used in over 40 projects at VUMC, including cancer clinical trial pre-screening, assessing genetic testing information, drug repurposing, streamlining clinical workflows, and more.
Same Day Discharge Plan for Pediatric Patients
Jamie Robinson developed a same day discharge plan for children undergoing gastrostomy tube placement. It decreased length of stay by over 24 hours for those patients compared to the traditional clinical practice guideline.
Telehealth for IBD Patients
Sara Horst has worked with VCLIC on several research and evaluation projects focused on improving care for patients with gastrointestinal conditions. These include a multi-site study with Columbia on outcomes in cirrhosis patients seen by gastroenterologists (to be presented at DDW 2025), analyses of telehealth utilization and patient-reported outcomes in inflammatory bowel disease, and an evaluation of eVisit effectiveness on patient outcomes.
Early Warning System for Adverse Outcomes
Brian Douthit has led and contributed to a range of impactful initiatives across VUMC, including supporting researchers with Epic queries to improve care for patients with early-stage CKD and teens with diabetes, preparing the CONCERN early warning system for implementation to reduce adverse outcomes, and advancing virtual nursing care models to enhance discharge education, patient throughput, and nursing workflow. He has also helped reduce documentation burden—most notably through work on pressure injury prevention and steering committee engagement—and is actively developing a toolkit to streamline nursing documentation. These efforts, supported by VCLIC's infrastructure and collaborative environment, have resulted in multiple presentations, publications, and sustained improvements to care delivery and operational efficiency.
Pre/Post-Op Order Sets
Parker Evans led the development and implementation of pediatric appendicitis pre- and post-operative order sets, aligning with their thesis project to improve clinical efficiency and guideline adherence. The postoperative order set supported a new departmental initiative to reduce unnecessary discharge antibiotic use, and early outcomes show high uptake of the preoperative set (~80% of patients) alongside a meaningful reduction in antibiotic prescribing without increases in complications.
Algorithmovigilance & Pharmacy Overview
Scott Nelson has led and supported a wide range of clinical, operational, and national initiatives spanning decision support, digital health, and pharmacy informatics. Highlights include leading VUMC's CAHIIM accreditation for MS-ACI, receiving the TODE Award, developing algorithm monitoring tools for AI model performance (VAMOS AI and algorithmovigilance), and optimizing Epic functionality—such as improving pharmacist-facing clinical overview reports, implementing CancelRx, and integrating KASPER and CSMD into workflows. He has advanced safety and stewardship through medication alerts (e.g., for ketorolac, valproic acid, and IV/PO steroid duplication), refined prescription renewal protocols, and contributed to multi-institutional initiatives related to pediatric sedation, adverse drug event monitoring, and nephrotoxic medication exposure surveillance.
Pediatric Informatics Initiatives
Dave Johnson has supported several impactful quality improvement initiatives across pediatric and adult care settings. Collaborations include a PHM fellow–led project to improve timely antiviral administration for children hospitalized with influenza through dynamic Epic builds (order sets and BPAs), reducing alert burden while improving guideline concordance. They are also working with pediatric service lines to enhance outcomes reported to USNWR by transforming quarterly metrics into real-time dashboards and control charts using Power BI and Easy SPC, which has enabled more proactive, data-driven improvement efforts. Additional collaborations include support for adult ED process improvements that increased timely ECG completion for chest pain patients, and partnership with Dr. Heather Kreth and Ashley Ried to reduce pediatric behavioral health admissions and LOS through integrated Epic tools and real-time data monitoring.
Optimize ED Workflows
Dario Giuse has led efforts to optimize ED workflows by developing six real-time dashboards that monitor operational bottlenecks—such as CT delays, inpatient bed availability, and discharge throughput—and trigger targeted alerts to ED staff, with nearly 100 alerts generated in four months. They have also advanced real-time clinical decision support through Word Cloud RTE mechanisms, enabling dynamic interventions like auto-consults for patients with insulin pumps, recruitment for research studies (e.g., DVT and fatty liver disease), and improved documentation of comorbidities via Care Everywhere data, ultimately supporting both clinical care and operational efficiency.
Improved NICU Processes
Eva Dye has led the integration and optimization of the NICU's participation in the Vermont Oxford Network (VON) registry, launching a new eStar-based system in 2024 and streamlining data capture through smartform automation, custom Workbench columns, and reporting tools that support quarterly QI outcomes. They have organized and documented legacy vNICU data in Databricks for research access, developed a REDCap request system that has fulfilled over a dozen NICU data requests, and are advising the development of a seamless database to connect NICU inpatient and developmental follow-up data for long-term outcomes tracking. Additional contributions include supporting NICU research through an AHRQ-funded RN workload study, serving as a key liaison for eStar and analytics enhancements, and providing frontline clinical staff with training and tools to improve their use of NICU-specific informatics resources.
Wael Alrifai led the successful integration of NICU clinical databases into VUMC systems, a complex project spanning smartform design to backend infrastructure using Databricks. The effort, supported by multiple VCLIC collaborators, including Eva Dye, resulted in significantly improved data accuracy, substantial efficiency gains equivalent to one RN FTE annually, and a reproducible framework now being scaled to support data operations for the Pediatric Heart Institute.
Healthcare via Text Message
Lindsay Mayberry has played a key role in advancing patient-centered, technology-enabled interventions for diabetes and social determinants of health. They serve as primary mentor to Cristin Fritz on her Techquity Challenge project, which is now launching an RCT of the Text Connect intervention for hospitalized families screening positive for food insecurity. They also lead the Engagement Core for the PCORI-funded PRECIDENTD study comparing diabetes medications for cardiovascular and kidney outcomes, and collaborate on multiple NIH- and AHRQ-funded initiatives—including RCTs and implementation studies—to evaluate and scale precision-medicine and patient portal interventions that support diabetes self-management and address social risk factors among patients with multiple chronic conditions.
Improved Medical Student Informatics Courses
Julian Genkins created and taught the first year of a longitudinal 4-year informatics curriculum for medical students at Vanderbilt—part of a reimagined health system science curriculum, Foundations of Professional Responsibility. Also, in his work with the healthy planet team, he deployed a combined colorectal cancer screening health maintenance topic which enables a one-stop shop for reviewing and acting on all colorectal cancer screening needs for patients.
Blood Product Tracker
Jennifer Andrews developed a blood tracker, published it, and did a podcast about it, and Epic is putting it into Foundation!
DiscoverE Program
Shelagh Mulvaney is the Principal Investigator leading Discovery Diabetes, an ADA-funded initiative aimed at reducing psychosocial barriers and improving self-care among adolescents and young adults with type 1 diabetes. The program integrates into existing clinic workflows including PROs administered through MyHealthatVanderbilt and extends that to include collaborative goal setting and support between visits via mobile personalized digital stories and SMS. An RCT to evaluate the intervention is planned, following feedback from patients and clinicians on the interface and workflow.
Pediatric Hem/Onc Informatics Impacts
Christine Smith has led and supported several informatics initiatives to improve care and documentation in pediatric hematology/oncology. She recently demonstrated how a standardized PHO-specific attestation smartphrase improved documentation accuracy and billing levels, with downstream financial and operational benefits. As a content expert, she contributed to the development of a real-time sickle cell dashboard for health maintenance tracking, collaborated on a patient- and provider-facing sickle cell pain action plan, and helped initiate the build of a PHO-specific discharge order set to support safe prescribing and anticipatory guidance at discharge.
Measurement Based Care for OUD
Thomas Reese is advancing clinical decision support and patient monitoring through two innovative digital health initiatives. He is implementing a SMART on FHIR app to reduce bleeding risk in patients on anticoagulants, as part of a multisite randomized trial with several resulting publications. In parallel, he is deploying a measurement-based care tool for opioid use disorder that captures patient-reported symptoms like cravings and side effects between visits, enabling near real-time response; a grant is in development to scale this approach in a multisite trial.
