Enhancing an EMR-Based Real-Time Sepsis Alert System Performance Through Machine Learning (Michigan)

Project Final Report (PDF, 433.19 KB) Disclaimer

Project Details - Ended

Project Categories

Summary:

Sepsis, a severe systemic infection with significant morbidity and mortality, hospitalizes over 1.6 million people in the United States annually at a cost exceeding $20 billion. Although the diagnostic criteria for sepsis appear relatively straightforward, the disease is diagnostically challenging and commonly underdetected. A major contributing factor to this is that many septic patients may initially present to the emergency department without organ dysfunction or evidence of shock and only develop these later during their stay, making early clinical recognition difficult.

Clinical decision support (CDS) tools have increasingly been used in the early identification of sepsis patients, but existing tools may miss 20 to 30 percent of sepsis patients and frequently misidentify nonsepsis patients as sepsis patients. Current tools have two specific limitations: (1) they lack a mechanism to learn from their past errors and consequently repeat the same mistakes and (2) the decisionmaking process is fixed and treats all patients in the same way despite their differences.

This project will use machine learning to enhance an existing sepsis CDS tool, Sepsis-Alert, to address these two limitations. This technology develops methods capable of learning from data and making predictions or decisions on its own. Examples of current clinical use include cancer prediction and prognosis, classification of depression patients and normal people, detection of Alzheimer’s disease, outbreak detection, and many others. The enhanced tool, to be called Intelligent Sepsis Alert, will use machine learning to recognize the subtleties of sepsis, categorize patients, and learn from its own mistakes to avoid repeating them.

The specific aims of this project are as follows:

  • Design Intelligent Sepsis Alert, based upon Sepsis-Alert 
  • Test and refine Intelligent Sepsis Alert and integrate a Real-Time Alert Module into the electronic medical record 
  • Operate Intelligent Sepsis Alert prospectively with cycles of online testing followed by offline learning and optimization 
  • Establish alert performance and clinical utility by applying the final real-time alert module to emergency department patients 

This project will deliver a highly accurate and advanced program readily adoptable by any health system or hospital to improve sepsis care and create a safer health care environment.

This project does not have any related annual summary.

Enhancing an EMR-Based Real-Time Sepsis Alert System Performance Through Machine Learning - Final Report

Citation:
Sherwin R. Enhancing an EMR-Based Real-Time Sepsis Alert System Performance Through Machine Learning - Final Report. (Prepared by Wayne State University under Grant No. R21 HS024750). Rockville, MD: Agency for Healthcare Research and Quality, 2019. (PDF, 433.19 KB)

The findings and conclusions in this document are those of the author(s), who are responsible for its content, and do not necessarily represent the views of AHRQ. No statement in this report should be construed as an official position of AHRQ or of the U.S. Department of Health and Human Services. (Persons using assistive technology may not be able to fully access information in this report. For assistance, please contact Corey Mackison)
Principal Investigator: 
Document Type: 
This project does not have any related resource.
This project does not have any related survey.
This project does not have any related project spotlight.
This project does not have any related survey.
This project does not have any related story.
This project does not have any related emerging lesson.