Dataset Information

A Machine Learning Algorithm to Predict Severe Sepsis and Septic Shock: Development, Implementation, and Impact on Clinical Practice.

ABSTRACT:

Objectives

Develop and implement a machine learning algorithm to predict severe sepsis and septic shock and evaluate the impact on clinical practice and patient outcomes.

Design

Retrospective cohort for algorithm derivation and validation, pre-post impact evaluation.

Setting

Tertiary teaching hospital system in Philadelphia, PA.

Patients

All non-ICU admissions; algorithm derivation July 2011 to June 2014 (n = 162,212); algorithm validation October to December 2015 (n = 10,448); silent versus alert comparison January 2016 to February 2017 (silent n = 22,280; alert n = 32,184).

Interventions

A random-forest classifier, derived and validated using electronic health record data, was deployed both silently and later with an alert to notify clinical teams of sepsis prediction.

Measurement and main result

Patients identified for training the algorithm were required to have International Classification of Diseases, 9th Edition codes for severe sepsis or septic shock and a positive blood culture during their hospital encounter with either a lactate greater than 2.2 mmol/L or a systolic blood pressure less than 90 mm Hg. The algorithm demonstrated a sensitivity of 26% and specificity of 98%, with a positive predictive value of 29% and positive likelihood ratio of 13. The alert resulted in a small statistically significant increase in lactate testing and IV fluid administration. There was no significant difference in mortality, discharge disposition, or transfer to ICU, although there was a reduction in time-to-ICU transfer.

Conclusions

Our machine learning algorithm can predict, with low sensitivity but high specificity, the impending occurrence of severe sepsis and septic shock. Algorithm-generated predictive alerts modestly impacted clinical measures. Next steps include describing clinical perception of this tool and optimizing algorithm design and delivery.

SUBMITTER: Giannini HM

PROVIDER: S-EPMC8635476 | biostudies-literature | 2019 Nov

REPOSITORIES: biostudies-literature

ACCESS DATA

Publications

A Machine Learning Algorithm to Predict Severe Sepsis and Septic Shock: Development, Implementation, and Impact on Clinical Practice.

Giannini Heather M HM Ginestra Jennifer C JC Chivers Corey C Draugelis Michael M Hanish Asaf A Schweickert William D WD Fuchs Barry D BD Meadows Laurie L Lynch Michael M Donnelly Patrick J PJ Pavan Kimberly K Fishman Neil O NO Hanson C William CW Umscheid Craig A CA

Critical care medicine 20191101 11

<h4>Objectives</h4>Develop and implement a machine learning algorithm to predict severe sepsis and septic shock and evaluate the impact on clinical practice and patient outcomes.<h4>Design</h4>Retrospective cohort for algorithm derivation and validation, pre-post impact evaluation.<h4>Setting</h4>Tertiary teaching hospital system in Philadelphia, PA.<h4>Patients</h4>All non-ICU admissions; algorithm derivation July 2011 to June 2014 (n = 162,212); algorithm validation October to December 2015 (n ...[more]

PMID: 31389839

Similar Datasets

Project description:BackgroundThe role of recombinant activated protein C (aPC) during sepsis is still controversial. It showed anti-inflammatory effect and improved the microvascular perfusion in experimental models of septic shock. The present study was aimed at testing the hypothesis that recombinant aPC therapy improves the microcirculation during severe sepsis.MethodsProspective observational study on patients admitted in a 12-beds intensive care unit of a university hospital from July 2010 to December 2011, with severe sepsis and at least two sepsis-induced organ failures occurring within 48 hours from the onset of sepsis, who received an infusion of aPC (24 mcg/kg/h for 96 hours) (aPC group). Patients with contraindications to aPC administration were also monitored (no-aPC group).At baseline (before starting aPC infusion, T0), after 24 hours (T1a), 48 hours (T1b), 72 hours (T1c) and 6 hours after the end of aPC infusion (T2), general clinical and hemodynamic parameters were collected and the sublingual microcirculation was evaluated with sidestream dark-field imaging. Total vessel density (TVD), perfused vessel density (PVD), De Backer score, microvascular flow index (MFIs), the proportion of perfused vessels (PPV) and the flow heterogeneity index (HI) were calculated for small vessels. The perfused boundary region (PBR) was measured as an index of glycocalyx damage. Variables were compared between time points and groups using non parametric or parametric statistical tests, as appropriate.ResultsIn the 13 aPC patients mean arterial pressure (MAP), base excess, lactate, PaO2/FiO2 and the Sequential Organ Failure Assessment (SOFA) score significantly improved over time, while CI and ITBVI did not change. MFIs, TVD, PVD, PPV significantly increased over time and the HI decreased (p < 0.05 in all cases), while the PBR did not change. No-aPC patients (n = 9) did not show any change in the microcirculation over time. A positive correlation was found between MFIs and MAP. TVD, PVD and De Backer score negatively correlated with norepinephrine dose, and the SOFA score negatively correlated with MFIs, TVD and PVD.ConclusionsaPC significantly improves the microcirculation in patients with severe sepsis/septic shock.Trial registrationNCT01806428.

Dataset Information

A Machine Learning Algorithm to Predict Severe Sepsis and Septic Shock: Development, Implementation, and Impact on Clinical Practice.

Objectives

Design

Setting

Patients

Interventions

Measurement and main result

Conclusions

Publications

A Machine Learning Algorithm to Predict Severe Sepsis and Septic Shock: Development, Implementation, and Impact on Clinical Practice.

Similar Datasets

OmicsDI is part of the ELIXIR infrastructure

Tweets