“…Our research suggests that both lower maximum and minimum albumin indicate an increased hospital mortality. This result is also supported by several previous studies [ 28 – 30 ]. As the most common protein in the human body, serum albumin has important physiological functions such as maintaining plasma osmotic pressure, buffering function and binding capacity.…”
Background
Early prediction of hospital mortality is crucial for ICU patients with sepsis. This study aimed to develop a novel blending machine learning (ML) model for hospital mortality prediction in ICU patients with sepsis.
Methods
Two ICU databases were employed: eICU Collaborative Research Database (eICU-CRD) and Medical Information Mart for Intensive Care III (MIMIC-III). All adult patients who fulfilled Sepsis-3 criteria were identified. Samples from eICU-CRD constituted training set and samples from MIMIC-III constituted test set. Stepwise logistic regression model was used for predictor selection. Blending ML model which integrated nine sorts of basic ML models was developed for hospital mortality prediction in ICU patients with sepsis. Model performance was evaluated by various measures related to discrimination or calibration.
Results
Twelve thousand five hundred fifty-eight patients from eICU-CRD were included as the training set, and 12,095 patients from MIMIC-III were included as the test set. Both the training set and the test set showed a hospital mortality of 17.9%. Maximum and minimum lactate, maximum and minimum albumin, minimum PaO2/FiO2 and age were important predictors identified by both random forest and extreme gradient boosting algorithm. Blending ML models based on corresponding set of predictors presented better discrimination than SAPS II (AUROC, 0.806 vs. 0.771; AUPRC 0.515 vs. 0.429) and SOFA (AUROC, 0.742 vs. 0.706; AUPRC 0.428 vs. 0.381) on the test set. In addition, calibration curves showed that blending ML models had better calibration than SAPS II.
Conclusions
The blending ML model is capable of integrating different sorts of basic ML models efficiently, and outperforms conventional severity scores in predicting hospital mortality among septic patients in ICU.
“…Our research suggests that both lower maximum and minimum albumin indicate an increased hospital mortality. This result is also supported by several previous studies [ 28 – 30 ]. As the most common protein in the human body, serum albumin has important physiological functions such as maintaining plasma osmotic pressure, buffering function and binding capacity.…”
Background
Early prediction of hospital mortality is crucial for ICU patients with sepsis. This study aimed to develop a novel blending machine learning (ML) model for hospital mortality prediction in ICU patients with sepsis.
Methods
Two ICU databases were employed: eICU Collaborative Research Database (eICU-CRD) and Medical Information Mart for Intensive Care III (MIMIC-III). All adult patients who fulfilled Sepsis-3 criteria were identified. Samples from eICU-CRD constituted training set and samples from MIMIC-III constituted test set. Stepwise logistic regression model was used for predictor selection. Blending ML model which integrated nine sorts of basic ML models was developed for hospital mortality prediction in ICU patients with sepsis. Model performance was evaluated by various measures related to discrimination or calibration.
Results
Twelve thousand five hundred fifty-eight patients from eICU-CRD were included as the training set, and 12,095 patients from MIMIC-III were included as the test set. Both the training set and the test set showed a hospital mortality of 17.9%. Maximum and minimum lactate, maximum and minimum albumin, minimum PaO2/FiO2 and age were important predictors identified by both random forest and extreme gradient boosting algorithm. Blending ML models based on corresponding set of predictors presented better discrimination than SAPS II (AUROC, 0.806 vs. 0.771; AUPRC 0.515 vs. 0.429) and SOFA (AUROC, 0.742 vs. 0.706; AUPRC 0.428 vs. 0.381) on the test set. In addition, calibration curves showed that blending ML models had better calibration than SAPS II.
Conclusions
The blending ML model is capable of integrating different sorts of basic ML models efficiently, and outperforms conventional severity scores in predicting hospital mortality among septic patients in ICU.
“…Serum albumin level is a common indicator for assessing a patient’s nutritional status, organ function, and comorbidity. The inflammatory state resulting from bacterial infection, which leads to the production of IL-1, TNF, and other cell mediators, can interfere with liver albumin synthesis, resulting in hypoalbuminemia [30]. There is currently a lack of literature on the relationship between mortality and serum albumin level in patients with E. coli infections.…”
Background
Escherichia coli
is one of the most common strains of extended-spectrum β-lactam (ESBL)-producing bacteria, and the prevention and treatment of ESBL-producing
E. coli
infections is an ongoing challenge. The clinical characteristics and outcomes of ESBL-producing
E. coli
bacteremia in non-transplant patients remain to be elucidated.
Methods
This retrospective study included 491 non-transplant patients with
E. coli
bloodstream infections (BSIs) from January 2013 to December 2016 and was conducted to investigate the risk factors, clinical features, and outcomes of these infections.
Results
Of the 491
E. coli
BSI patients, 57.6% suffered from infections with ESBL-producing strains. A multivariate analysis showed that urinary tract infection, prior use of cephalosporin, and treatment with β-lactam-β-lactamase inhibitor (BLBLI) combination antibiotics were independent risk factors for the development of ESBL-producing
E. coli
BSIs. The overall mortality rate in
E. coli
BSI patients was 14.46%, and there was no significant difference in the 28 day mortality rate between ESBL-producing
E. coli
and non-ESBL-producing
E. coli
BSI patients (14.8% vs. 14.0%, respectively;
P
= 0.953). Similarly, there was no difference between the community-acquired infection group and the nosocomial infection group. Hepatobiliary disease, carbapenem exposure, high APACHE II score, and hypoproteinemia were independent risk factors for death in
E. coli
BSI patients. Multivariate analysis showed that hypoproteinemia and severe disease were independent risk factors for death from
ESBL
-producing
E. coli
BSIs. Furthermore, there was no significant difference in the 28 day mortality between patients with
ESBL
-producing
E. coli
BSIs treated with carbapenem monotherapy versus those treated with BLBLI combination antibiotics (12.8% vs. 17.9%, respectively;
P
= 0.384).
Conclusions
Prior use of cephalosporin or BLBLI combination antibiotics increased the risk ratio for
ESBL
-producing
E. coli
infection. Hypoproteinemia and severe disease are independent risk factors for death in patients with
E. coli
BSIs. There was no significant difference in the 28 day prognosis of patients with ESBL-producing
E. coli
and those with non-ESBL-producing
E. coli
BSIs. These data do not support the conclusion that carbapenems might be more effective than BLBLI antibiotics for treatment of patients with BSIs cau...
“…This means that lower albumin levels correlate with severe systemic inflammation and organ failure [14]. Moreover, several studies demonstrated that low albumin levels correlated with adverse clinical outcomes [11, 15].…”
There has been no study exploring the prognostic values of neutrophil percentage-to-albumin ratio (NPAR). We hypothesised that NPAR is a novel marker of inflammation and is associated with all-cause mortality in patients with severe sepsis or septic shock. Patient data were extracted from the MIMIC-III V1.4 database. Only the data for the first intensive care unit (ICU) admission of each patient were used and baseline data were extracted within 24 h after ICU admission. The clinical endpoints were 30-, 90- and 365-day all-cause mortality in critically ill patients with severe sepsis or septic shock. Cox proportional hazards models and subgroup analyses were used to determine the relationship between NPAR and these clinical endpoints. A total of 2166 patients were eligible for this analysis. In multivariate analysis, after adjustments for age, ethnicity and gender, higher NPAR was associated with increased risk of 30-, 90- and 365-day all-cause mortality in critically ill patients with severe sepsis or septic shock. Furthermore, after adjusting for more confounding factors, higher NPAR remained a significant predictor of all-cause mortality (tertile 3 vs. tertile 1: HR, 95% CI: 1.29, 1.04–1.61; 1.41, 1.16–1.72; 1.44, 1.21–1.71). A similar trend was observed in NPAR levels stratified by quartiles. Higher NPAR was associated with increased risk of all-cause mortality in critically ill patients with severe sepsis or septic shock.
scite is a Brooklyn-based organization that helps researchers better discover and understand research articles through Smart Citations–citations that display the context of the citation and describe whether the article provides supporting or contrasting evidence. scite is used by students and researchers from around the world and is funded in part by the National Science Foundation and the National Institute on Drug Abuse of the National Institutes of Health.