Right dose, right now: using big data to optimize antibiotic dosing in the critically ill

Elbers, Paul; Girbes, Armand R.J.; Malbrain, Manu L N G; Bosman, Rob J.

doi:10.5603/ait.a2015.0061

Cited by 24 publications

(26 citation statements)

References 18 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…Secondly, and perhaps more importantly, for most predictive models formal evaluation or bed side implementation is currently lacking [38]. Based on our previous experience, developing a bed side decision support tool requires designing a model, pipeline and software with clinical implementation in mind [39,40]. For predictive models, this involves close collaboration between intensivists and data scientists for extensive feature engineering with a focus on features that are available in realtime, innovative approaches with respect to interpretability, actionable insights and feature importance, as well as extensive performance evaluations and impact analyses.…”

Section: Discussionmentioning

confidence: 99%

Developing a Machine Learning prediction model for bedside decision support by predicting readmission or death following discharge from the Intensive Care unit

Thoral

Fornasa²,

Bruin³

et al. 2020

Preprint

Self Cite

View full text Add to dashboard Cite

Background Unexpected ICU readmission is associated with longer length of stay and an increase in mortality. Real time support systems could prevent untimely discharge from the ICU. We aim to develop a machine learning model for implementation at the bedside by predicting the risk of ICU readmission or death at time of potential discharge, showing feature importance and visualizing day-to-day changes in risk. Methods Data from adult patients, admitted to our mixed surgical-medical ICU between 2004 and 2016, were used in the analysis. Patient characteristics, clinical observations, (automated) physiological measurements, laboratory studies and treatment data were considered as model features. Different supervised learning algorithms were trained to predict ICU readmission and/or death, both within 7 days from ICU discharge, using 10-fold cross-validation. Feature importance was determined using SHapley Additive exPlanations. We constructed readmission probability-time curves to identify subgroups. Results Our dataset included 14,105 admissions. The combined readmission/mortality rate within seven days of ICU discharge was 5.3%. Using Gradient Boosting, the model achieved a Receiver Operating Characteristic AUC of 0.802 (95% CI 0.789-0.816) and a Precision-Recall AUC of 0.198 (95% CI 0.185-0.211). The most predictive features were well-known parameters, including physiological parameters, as well as less apparent features like nutritional support. Impact analysis using probability-time curves identified specific patients groups, that might lead to a change in discharge management with a relative risk reduction of 17%. Conclusions We developed a model that can accurately predict readmission and mortality after ICU discharge. Impact analysis showed that a relative risk reduction of 17% could be achievable. Given the large and increasing number of ICU admissions worldwide, this modest reduction may have significant impact for patients and society.

show abstract

Section: Discussionmentioning

confidence: 99%

Developing a Machine Learning prediction model for bedside decision support by predicting readmission or death following discharge from the Intensive Care unit

Thoral

Fornasa²,

Bruin³

et al. 2020

Preprint

Self Cite

View full text Add to dashboard Cite

show abstract

“…Fluid balance as a predictor of drug Vd available at the bedside would help critically ill ICU patients, who require adequate dosing within 24 h after the start of treatment and /or next dose adaptation 35 .…”

Section: Limitationmentioning

confidence: 99%

Covariate determinants of effective dosing regimens for time-dependent beta-lactam antibiotics for critically ill patients

Kaška¹,

Havel²,

Selke-Krulichova³

et al. 2018

Biomed Pap Med Fac Univ Palacky Olomouc Czech Repub

View full text Add to dashboard Cite

Aims. Critically ill patients undergoing aggressive fluid resuscitation and treated empirically with hydrosoluble timedependent beta-lactam antibiotics are at risk for sub-therapeutic plasma concentrations. The aim of this study was to assess the impact of two covariates -creatinine clearance (Cl cr ) and cumulative fluid balance (CFB) on pharmacokinetics/pharmacodynamics (PK/PD) target attainment within a week of treatment with meropenem (ME) or piperacillin/ tazobactam (PIP/TZB). Methods. In this prospective observational pharmacokinetic (PK) study, 18 critically ill patients admitted to a surgical Intensive Care Unit (ICU) were enrolled. The primary PK/PD target was free antibiotic concentrations above MIC at 100% of the dosing interval (100%fT>MIC) to obtain maximum bactericidal activity. Drug concentration was measured using liquid chromatography-tandem mass spectrometry. Results. The treatment of both 8 septic patients with IV extended ME dosing 2 g/3 h q8 h and 10 polytraumatized patients with IV intermittent PIP/TZB dosing 4.0/0.5 g q8 h was monitored. 8/18 patients (44%) manifested augmented renal clearence (ARC) where Cl cr ≥130 mL/min/1.73m 2 . Maximum changes were reported on days 2-3: the median positive CFB followed by the large median volume of distribution: Vd me =70.3 L (41.9-101.5), Vd pip = 46.8 L (39.7-60.0). 100%fT me >MIC was achieved in all patients on ME (aged ≥60 years), and only in two patients (non-ARC, aged ≥65 years) out of 10 on PIP/TZB. A mixed model analysis revealed positive relationship of CFB pip with Vd pip (P=0.021). Conclusion. Assuming that the positive correlation between CFB and Vd exists for piperacillin in the setting of the pathological state, then CFB should predict Vd pip across subjects at each and every time point.

show abstract

“…However, implementation of decision support systems is a complex process that requires a true multidisciplinary endeavor. And although several dashboards have been described and prospectively evaluated, much work remains to be done to accomplish interfacing of clinical and PK/PD dashboard systems with Electronic Health Records . A Learning Health System, such as the ICN Network described here, represents an ideal environment to facilitate such implementation, as it already harnesses the power of the electronic medical record, engages patients, families, and clinicians, and fosters research of novel interventions across participating institutions that ultimately will inform best practices.…”

Section: Vision For the Futurementioning

confidence: 99%

Learning Health Systems as Facilitators of Precision Medicine

Ramsey

Mizuno

Vinks

et al. 2017

Clin Pharma and Therapeutics

View full text Add to dashboard Cite

Right dose, right now: using big data to optimize antibiotic dosing in the critically ill

Cited by 24 publications

References 18 publications

Developing a Machine Learning prediction model for bedside decision support by predicting readmission or death following discharge from the Intensive Care unit

Developing a Machine Learning prediction model for bedside decision support by predicting readmission or death following discharge from the Intensive Care unit

Covariate determinants of effective dosing regimens for time-dependent beta-lactam antibiotics for critically ill patients

Learning Health Systems as Facilitators of Precision Medicine

Contact Info

Product

Resources

About