Impact of a Machine Learning-Driven SmartAlert on Repetitive Inpatient Lab Ordering

Presentation Time: 01:00 PM - 01:15 PM

Abstract Keywords: Artificial Intelligence/Machine Learning, Learning Health System, Adaptive Clinical Decision Support
Primary Track: AI and Care Outcomes
Programmatic Theme: Clinical Decision Support and Analytics

A machine learning-driven interruptive alert was implemented at our institution to encourage cessation of medically unnecessary inpatient lab tests. The alert was deployed in a randomized manner to study the prospective performance of the underlying lab stability prediction ML algorithm as well as the impact of the alert on clinician lab ordering behavior. The ML algorithm had PPV of 87% and CBC results within 28 hours of alert firing decreased from 66% to 47%.

Speaker:
April Liang, MD
Stanford University

Authors:
Fatemeh Amrollahi, PhD - Stanford University; Stephen Ma, MD, PhD - Stanford University School of Medicine; Yixing Jiang, PhD candidate - Stanford; Aakash Acharya, - - Stanford Health Care; Sreedevi Mony, - - Stanford Health Care; Soumya Punnathanam, - - Stanford Health Care; John McKeown, - - Stanford Health Care; Conor Corbin; Grace Kim, MS - Johns Hopkins University School of Medicine; David Mui, MD, MBA - Stanford University School of Medicine; Trevor Crowell, BA - Healthcare AI Applied Research Team, Stanford University; Tho Pham; David Svec, MD, MBA - Stanford Health Care; Lisa Shieh, MD, PhD - Stanford University School of Medicine; Christopher Sharp, MD - Stanford University School of Medicine; Jonathan Chen, MD, PhD - Stanford University Hospital;

---

Evaluating the Impact of Machine Learning Integration on EHR Workflows: A Study on Clinician Efficiency, Effectiveness, and Satisfaction

2025 Clinical Informatics Conference 25x5 Presentation

Presentation Time: 01:15 PM - 01:30 PM

Abstract Keywords: Artificial Intelligence/Machine Learning, Usability and Measuring User Experience, Human Factors Testing
Primary Track: AI and Care Outcomes
Programmatic Theme: Usability, Efficiency, and User Experience

This study examines the impact of integrating the eCART tool into EHR workflows on nurses’ and residents’ efficiency, effectiveness, and satisfaction during patient prioritization tasks. Results showed no significant difference in task time but increased confidence in decision-making with eCART. Nurses exhibited wide variability in task times, suggesting a need for targeted training. Residents demonstrated improved prioritization accuracy with eCART.

Speaker:
Ann Wieben, PhD, BSN RN NI-BC
University of Wisconsin Madison

Authors:
Ann Wieben, PhD, BSN RN NI-BC - University of Wisconsin Madison; Apoorva Maru, B.S./B.A. - University of Wisconsin-Madison; Joseph Reid, MSN, RN - AgileMD, Inc.; Dana Edelson, MD, MS - AgileMD, Inc.; Brian Patterson, MD MPH - University of Wisconsin-Madison;

---

Optimizing Nursing EHR Data Standardization with Cutting-Edge Prompt Engineering and RAG Pipeline

2025 Clinical Informatics Conference 25x5 Presentation

Presentation Time: 01:30 PM - 01:45 PM

Abstract Keywords: Bridging Analytics, Bedside Care, Clinical Documentation, and Education, Artificial Intelligence/Machine Learning, EHR Implementation and Optimization, Documentation Burden
Primary Track: AI and Care Outcomes
Programmatic Theme: Informatics-Driven Value-Based Healthcare

Nurses play an essential role in healthcare. However, nursing care is rarely examined due to limited availability of standardized nursing data. We tested the performance of prompt engineering techniques and large language models (LLMs) in a retrieval-augmented generation pipeline in standardizing local nursing care plan terms from electronic health records. Our results show that while LLMs are not yet reliable for full automation of data standardization, they can reduce human workload and streamline the process.

Speaker:
Vedant Upganlawar, MS in Computer Science
University of Florida

Authors:
Yingwei Yao, PHD - University of Florida; Gail Keenan, PHD, RN - University of Florida; Tamara Macieira, PhD, RN - University of Florida;

---

Discrete Event Simulation Interventions In ICU Settings To Determine Staffing and Resource Allocation Scenarios to Improve Patient Length Of Stay

Presentation Time: 01:45 PM - 02:00 PM

Abstract Keywords: Data Science, Interprofessional Collaboration, Workflow Efficiency
Primary Track: AI and Care Outcomes
Programmatic Theme: Informatics-Driven Value-Based Healthcare

Increased ICU length of stay (LOS) is correlated with higher patient mortality, while lower patient-to-nurse ratios can reduce LOS. This study proposes a discrete-time event model, using patient care activities recorded in ICU data to chart patient paths in trauma ICU settings. The model aims to identify staffing bottlenecks and quantify the marginal impact of increased staffing on LOS and process wait times, providing insights for optimizing ICU workflows and improving patient outcomes.

Speaker:
Patrick Wedgeworth, MD, MISM
University of Washington

Authors:
Christophe Combemale, PhD - Carnegie Mellon University; Dustin Ferrone, MSE - Valdos Consulting; Shantanu Samant, MS - Valdos Consulting; Keerthana Sattiraju, Student - University of Washington;

---