A Novel Approach to Sepsis Detection Using Big Data
Timely initiation of adequate antimicrobial therapy saves patient lives while waiving of antibiotic treatment, when no infection is present, safeguards antibiotic effectiveness. Yet early and reliable diagnosis of sepsis in the critically ill as a basis for this important treatment decision remains a major clinical challenge. In the December 2016 issue of Critical Care Medicine, SCIDATOS alliance researchers report on a computerized algorithm that uses data from the electronic medical records (EMRs) of the surgical intensive care unit (ICU) at the University Hospital Mannheim to automatically capture the Systemic Inflammatory Response Syndrome (SIRS) - an important feature of sepsis - in individual patients.
The algorithm determines the number of SIRS criteria met for each minute. Based on this, intuitive measures of SIRS dynamics for a time period of interest were defined. These so-called SIRS-descriptors are (1) the average number of SIRS-criteria met each minute of the period, (2) the first-to-last minute difference in SIRS-criteria for trend, and (3) the count of minute-to-minute changes in SIRS criteria reflecting fluctuation over the period. The SIRS descriptors were applied in polytrauma patients identified from the EMR. The average number of SIRS-criteria over the first 24 hours following ICU admission outperformed the classical static definition of SIRS in predicting sepsis in polytrauma patients. Combining average and first-to-last minute difference for the 24 hour-period prior to the time of clinical sepsis diagnosis in a nested case-control analysis rivaled current biomarkers in discriminatory power for sepsis diagnosis.
This approach to early recognition of deterioration in a patient’s state is a step toward improved electronic sepsis surveillance, and further steps in this direction are currently taken by the SCIDATOS research alliance.
Lindner HA, Balaban Ü, Sturm T, Weiß C, Thiel M, Schneider-Lindner V. 2016. An Algorithm for Systemic Inflammatory Response Syndrome Criteria-Based Prediction of Sepsis in a Polytrauma Cohort. Crit Care Med. 44(12):2199-2207.
Dellinger RP. 2016. Dynamic Systemic Inflammatory Response Syndrome Monitoring: A reSIRSgence? Crit Care Med. 44(12):2285-2286.
Namas RA, Vodovotz Y. 2016. From static to dynamic: a sepsis-specific dynamic model from clinical criteria in polytrauma patients. Ann Transl Med. 4(24):492. doi: 10.21037/atm.2016.11.72.
Nandi U, Puskarich MA, Jones AE. 2016. Big data to the rescue of systemic inflammatory response syndrome: is electronic data mining the way of the future? Ann Transl Med. 4(23):465. doi: 10.21037/atm.2016.11.55.
Lindner HA, Thiel M, Schneider-Lindner V. 2017. Automated dynamic sepsis surveillance with routine data: opportunities and challenges. Ann Transl Med. 5(3):66. doi: 10.21037/atm.2016.11.55.