Early Machine-Human Interface around Sepsis Severity Identification: From Diagnosis to Improved Management?

Area of Science:

• Emergency Medicine
• Critical Care Medicine
• Health Informatics

Background:

• Sepsis remains a leading cause of mortality, necessitating early and accurate identification.
• Current diagnostic methods can be time-consuming, delaying critical interventions.
• The Surviving Sepsis Campaign (SSC) guidelines emphasize timely resuscitation and management.

Purpose of the Study:

• To evaluate the performance of a computerized sepsis sniffer algorithm (CSSA) for early sepsis detection in the emergency department (ED).
• To assess the effectiveness of human decision-making in activating a multidisciplinary sepsis and shock response team (SSRT).
• To analyze the combined performance of an automated tool and human oversight in sepsis management.

Main Methods:

• Prospective and historical cohort study design.
• Evaluation of a computerized sepsis sniffer algorithm (CSSA) for sepsis identification.
• Analysis of the activation patterns and appropriateness of a multidisciplinary sepsis and shock response team (SSRT).

Main Results:

• The CSSA achieved 100% sensitivity and 96.2% specificity for identifying sepsis in the ED.
• The SSRT was appropriately activated in 34.1% of severe sepsis or septic shock cases.
• In 53% of cases meeting SSRT criteria, the critical care team was consulted instead of activating the SSRT.

Conclusions:

• A two-step approach combining a sensitive electronic tool (CSSA) with human decision-making (SSRT activation) is effective for sepsis management.
• The CSSA shows high accuracy for early sepsis detection.
• Human oversight refines the application of critical care resources for sepsis patients.