Machine Learning-Based Prediction of Sepsis Onset in ICU Patients: A Multicenter Prospective Study
Kevin L. Zhao, PhD; Patricia M. O'Brien, MD, PhD; Steven J. Anderson, MD; Rachel K. Patel, MS
Department of Biomedical Informatics, Columbia University, New York, NY, USA; Division of Critical Care Medicine, NewYork-Presbyterian Hospital, New York, NY, USA
Background: Early detection of sepsis remains a critical challenge in intensive care. This study develops and validates a machine learning model for predicting sepsis onset up to 6 hours before clinical recognition.
Methods: We collected data from 15,847 ICU admissions across five tertiary care centers. Features included vital signs, laboratory values, demographics, and clinical interventions recorded at hourly intervals.
Results: The gradient boosting model achieved an area under the receiver operating characteristic curve (AUROC) of 0.94 (95% CI: 0.92-0.96) for sepsis prediction 6 hours before onset, with sensitivity of 87.3% and specificity of 91.2%.
Conclusions: Our machine learning model demonstrates high accuracy for early sepsis prediction and has the potential to significantly improve patient outcomes through timely intervention.
Sepsis is a life-threatening organ dysfunction caused by a dysregulated host response to infection. Despite advances in critical care, sepsis remains associated with mortality rates of 25-30% in ICU settings.
References
1. Singer M, Deutschman CS, Seymour CW, et al. The Third International Consensus Definitions for Sepsis and Septic Shock (Sepsis-3). JAMA. 2016;315(8):801-810.
How to Cite
APA Style
Kevin L. Zhao, PhD; Patricia M. O'Brien, MD, PhD; Steven J. Anderson, MD; Rachel K. Patel, MS (2025). Machine Learning-Based Prediction of Sepsis Onset in ICU Patients: A Multicenter Prospective Study. American Journal of Advanced Medical and Surgical Sciences, 2(1), 43-58. https://doi.org/10.XXXXX/ajoams.2025.0104
Vancouver Style
Kevin L. Zhao, PhD; Patricia M. O'Brien, MD, PhD; Steven J. Anderson, MD; Rachel K. Patel, MS. Machine Learning-Based Prediction of Sepsis Onset in ICU Patients: A Multicenter Prospective Study. Am J Adv Med Surg Sci. 2025;2(1):43-58. doi:10.XXXXX/ajoams.2025.0104
License: This article is published under the CC BY 4.0 license.
Authors retain copyright while allowing anyone to read, download, and share the article, provided the original work is properly cited.