Novel Machine Learning on Asynchronous Clinical Pages to Predict Clinical Deterioration

Poster Number: P49
Presentation Time: 05:00 PM - 06:30 PM

Abstract Keywords: Machine Learning, Workflow, Clinical Decision Support, Natural Language Processing, Critical Care, Diagnostic Systems, Nursing Informatics
Primary Track: Applications
Programmatic Theme: Clinical Informatics

Current early warning scores for predicting patient deterioration rely on manual EHR data entry around a few limited parameters. We propose a novel machine learning approach using clinical pages, which provide insight into clinicians’ intuition and already part of clinical workflow, to predict deterioration events. Our model detected 53% of events within 24 hours at a specificity of 0.80, outperforming existing methods. This approach can improve patient care, reduce nursing workload, and enhance clinical decision-making.

Speaker(s):
Isabel Arvelo, BA
Vanderbilt University

Author(s):
Bryan Steitz, PhD - Vanderbilt University Medical Center; Adam Wright, PhD - Vanderbilt University Medical Center; Kipp Shipley, DNP - Vanderbilt University Medical Center;