Dataset Information

Developing well-calibrated illness severity scores for decision support in the critically ill.

ABSTRACT: Illness severity scores are regularly employed for quality improvement and benchmarking in the intensive care unit, but poor generalization performance, particularly with respect to probability calibration, has limited their use for decision support. These models tend to perform worse in patients at a high risk for mortality. We hypothesized that a sequential modeling approach wherein an initial regression model assigns risk and all patients deemed high risk then have their risk quantified by a second, high-risk-specific, regression model would result in a model with superior calibration across the risk spectrum. We compared this approach to a logistic regression model and a sophisticated machine learning approach, the gradient boosting machine. The sequential approach did not have an effect on the receiver operating characteristic curve or the precision-recall curve but resulted in improved reliability curves. The gradient boosting machine achieved a small improvement in discrimination performance and was similarly calibrated to the sequential models.

SUBMITTER: Cosgriff CV

PROVIDER: S-EPMC6695410 | biostudies-literature | 2019

REPOSITORIES: biostudies-literature

ACCESS DATA

Publications

Developing well-calibrated illness severity scores for decision support in the critically ill.

Cosgriff Christopher V CV Celi Leo Anthony LA Ko Stephanie S Sundaresan Tejas T Armengol de la Hoz Miguel Ángel MÁ Kaufman Aaron Russell AR Stone David J DJ Badawi Omar O Deliberato Rodrigo Octavio RO

NPJ digital medicine 20190815

Illness severity scores are regularly employed for quality improvement and benchmarking in the intensive care unit, but poor generalization performance, particularly with respect to probability calibration, has limited their use for decision support. These models tend to perform worse in patients at a high risk for mortality. We hypothesized that a sequential modeling approach wherein an initial regression model assigns risk and all patients deemed <i>high risk</i> then have their risk quantifie ...[more]

PMID: 31428687

Similar Datasets

Project description:Plasma cell-free DNA (cfDNA) levels have been associated with disease and survival status in septic humans and dogs. To date, studies investigating cfDNA levels in association with critical illness in foals are lacking. We hypothesized that cfDNA would be detectable in the plasma of foals, that septic and sick-nonseptic foals would have significantly higher cfDNA levels compared to healthy foals, and that increased cfDNA levels would be associated with non-survival. Animals used include 80 foals of 10 days of age or less admitted to a tertiary referral center between January and July, 2020 were stratified into three categories: healthy (n = 34), sick non-septic (n = 11) and septic (n = 35) based on specific criteria. This was a prospective clinical study. Blood was collected from critically ill foals at admission or born in hospital for cfDNA quantification and blood culture. Previously published sepsis score (SS) and neonatal SIRS score (NSIRS) were also calculated. SS, NSIRS, blood culture status and cfDNA concentrations were evaluated to predict survival. Continuous variables between groups were compared using Kruskal-Wallis ANOVA with Dunn's post hoc test. Comparisons between two groups were assessed using the Mann-Whitney U-test or Spearman rank for correlations. The performance of cfDNA, sepsis score and NSIRS score to predict survival was assessed by receiver operator characteristic (ROC) curve analysis including area under the curve, sensitivity and specificity using cutoffs. Plasma cfDNA was detectable in all foals. No significant differences in cfDNA concentration were detected between healthy foals and septic foals (P = 0.65) or healthy foals and sick non-septic foals (P = 0.88). There was no significant association between cfDNA and culture status, SS, NSIRS or foal survival. SS (AUC 0.85) and NSIRS (AUC 0.83) were superior to cfDNA (AUC 0.64) in predicting survival. Although cfDNA was detectable in foal plasma, it offers negligible utility to diagnose sepsis or predict survival in critical illness in neonates.

Dataset Information

Developing well-calibrated illness severity scores for decision support in the critically ill.

Publications

Developing well-calibrated illness severity scores for decision support in the critically ill.

Similar Datasets

OmicsDI is part of the ELIXIR infrastructure

Tweets