Collaborative Research Database Research Articles

Development and External Validation of a Nomogram for Predicting Acute Kidney Injury in Cardiogenic Shock Patients in Intensive Care Unit.

BackgroundThe aim of this study was to construct and external validate a nomogram for predicting cardiogenic shock acute kidney injury (CS-AKI) in patients in intensive care unit (ICU).MethodsAll patients diagnosed with CS from the Medical Information Mart for Intensive Care IV (MIMIC-IV) database and the eICU Collaborative Research Database (eICU-CRD) were included in this study. Least absolute shrinkage and selection operator (LASSO) regression and recursive feature elimination for support vector machine (SVM-RFE) were used to determine the overlapping clinical features associated with CS-AKI. The predictive nomogram was established based on the significant clinical parameters and externally verified in this study.ResultsLASSO regression and SVM-RFE demonstrated that Charlson Comorbidity Index (CCI), usage of mechanical ventilation, SOFA score, white blood cell, albumin, eGFR, anion gap, and positive fluid balance were closely associated with CS-AKI in the training cohort. The predictive nomogram based on the eight parameters showed good predictive performance as calculated by C-index were 0.823 (95% confidence index, 95% CI 0.798–0.849), 0.819 (95% CI 0.769–0.849), and 0.733 (95% CI 0.704–0.763) in the training set, in the internal validation set and in the external validation sets, respectively. Moreover, the nomogram exhibited not only encouraging calibration ability but also great clinical utility in the training set and in the validation sets.ConclusionCCI, usage of mechanical ventilation, SOFA score, white blood cell, albumin, eGFR, anion gap, and positive fluid balance were closely associated with CS-AKI. The predictive nomogram for CS-AKI manifested well-predictive ability for the identification of ICU patients with CS-AKI.

Association between platelets and in-hospital mortality in critically ill patients with tumours: a retrospective cohort study

ObjectivesPlatelet count is an independent predictor of mortality in patients with cancer. It remains unknown whether the platelet count is related to in-hospital mortality in severely ill patients with tumours.DesignA...

The Global Open Source Severity of Illness Score (GOSSIS).

To develop and demonstrate the feasibility of a Global Open Source Severity of Illness Score (GOSSIS)-1 for critical care patients, which generalizes across healthcare systems and countries. A merger of several critical care multicenter cohorts derived from registry and electronic health record data. Data were split into training (70%) and test (30%) sets, using each set exclusively for development and evaluation, respectively. Missing data were imputed when not available. Two large multicenter datasets from Australia and New Zealand (Australian and New Zealand Intensive Care Society Adult Patient Database [ANZICS-APD]) and the United States (eICU Collaborative Research Database [eICU-CRD]) representing 249,229 and 131,051 patients, respectively. ANZICS-APD and eICU-CRD contributed data from 162 and 204 hospitals, respectively. The cohort included all ICU admissions discharged in 2014-2015, excluding patients less than 16 years old, admissions less than 6 hours, and those with a previous ICU stay. Not applicable. GOSSIS-1 uses data collected during the ICU stay's first 24 hours, including extrema values for vital signs and laboratory results, admission diagnosis, the Glasgow Coma Scale, chronic comorbidities, and admission/demographic variables. The datasets showed significant variation in admission-related variables, case-mix, and average physiologic state. Despite this heterogeneity, test set discrimination of GOSSIS-1 was high (area under the receiver operator characteristic curve [AUROC], 0.918; 95% CI, 0.915-0.921) and calibration was excellent (standardized mortality ratio [SMR], 0.986; 95% CI, 0.966-1.005; Brier score, 0.050). Performance was held within ANZICS-APD (AUROC, 0.925; SMR, 0.982; Brier score, 0.047) and eICU-CRD (AUROC, 0.904; SMR, 0.992; Brier score, 0.055). Compared with GOSSIS-1, Acute Physiology and Chronic Health Evaluation (APACHE)-IIIj (ANZICS-APD) and APACHE-IVa (eICU-CRD), had worse discrimination with AUROCs of 0.904 and 0.869, and poorer calibration with SMRs of 0.594 and 0.770, and Brier scores of 0.059 and 0.063, respectively. GOSSIS-1 is a modern, free, open-source inhospital mortality prediction algorithm for critical care patients, achieving excellent discrimination and calibration across three countries.

Serum potassium level, variability and in-hospital mortality in acute myocardial infarction.

Clinical guidelines recommend an optimal serum potassium concentration between 4.0 and 5.0mmol/L in patients with acute myocardial infarction (AMI), which was based on lower-quality evidence from more than 20years ago. Therefore, it is essential to re-evaluate the range of optimal potassium levels in patients with AMI in intensive care unit (ICU). This was a retrospective study based on Philips eICU Collaborative Research Database, which covered 9776 patients with AMI between 2014 and 2015. All patients had more than or equal to 2 serum potassium measurements and were categorized by the mean serum potassium level (<3.5, 3.5-4.5, 4.5-5.5, ≥5.5mmol/L) and potassium variability (1st, 2nd, and ≥3rd standard deviation (SD)). Binary logistic regression was used to determine the association between mean potassium levels, variability and in-hospital mortality in AMI. Of all 9776 AMI patients in ICU, 8731 (89.3%) patients were included. A total of 69847 potassium measurements were performed in these patients. There was a J-shaped relationship between mean serum potassium level and in-hospital mortality. The lowest mortality (mortality rate, 7.2%; 95% CI, 6.57%-7.76%) was observed in patients with mean potassium level between 3.5 and 4.5mmol/L and a low potassium variability within the 1st SD. Logistic regression showed that the risk of in-hospital mortality is highest when the mean potassium level ≥5.5mmol/L (57.6%; 95% Cl, 45.02%-70.24%; multivariable adjusted OR, 14.8; 95% CI, 8.4-26.2) compared to the reference group of 3.5-4.5mmol/L and potassium variability within the 3rd SD (16.5%; 95% Cl, 15.19%-17.88%; multivariable adjusted OR, 3.3; 95% CI, 2.7-4.1) compared to 1st SD. Several sensitivity analyses confirmed these results. Among AMI patients in ICU, the minimum risk of in-hospital mortality was observed in those with mean potassium levels between 3.5 and 4.5mmol/L or a minimal potassium variability compared to those who had higher or lower values.

A machine learning-based risk stratification tool for in-hospital mortality of intensive care unit patients with heart failure

BackgroundPredicting hospital mortality risk is essential for the care of heart failure patients, especially for those in intensive care units.MethodsUsing a novel machine learning algorithm, we constructed a risk stratification tool that correlated patients’ clinical features and in-hospital mortality. We used the extreme gradient boosting algorithm to generate a model predicting the mortality risk of heart failure patients in the intensive care unit in the derivation dataset of 5676 patients from the Medical Information Mart for Intensive Care III database. The logistic regression model and a common risk score for mortality were used for comparison. The eICU Collaborative Research Database dataset was used for external validation.ResultsThe performance of the machine learning model was superior to that of conventional risk predictive methods, with the area under curve 0.831 (95% CI 0.820–0.843) and acceptable calibration. In external validation, the model had an area under the curve of 0.809 (95% CI 0.805–0.814). Risk stratification through the model was specific when the hospital mortality was very low, low, moderate, high, and very high (2.0%, 10.2%, 11.5%, 21.2% and 56.2%, respectively). The decision curve analysis verified that the machine learning model is the best clinically valuable in predicting mortality risk.ConclusionUsing readily available clinical data in the intensive care unit, we built a machine learning-based mortality risk tool with prediction accuracy superior to that of linear regression model and common risk scores. The risk tool may support clinicians in assessing individual patients and making individualized treatment.

Management of intoxicated patients – a descriptive outcome analysis of 4,267 ICU patients

IntroductionIntoxications are common in intensive care units (ICUs). The number of causative substances is large, mortality usually low. This retrospective cohort study aims to characterize differences of intoxicated compared to general ICU patients, point out variations according to causative agents, as well as to highlight differences between survivors and non-survivors among intoxicated individuals in a large-scale multi-center analysis.MethodsA total of 105,998 general ICU patients and 4,267 individuals with the admission diagnoses “overdose” and “drug toxicity” from the years 2014 and 2015 where included from the eICU Collaborative Research Database. In addition to comparing these groups with respect to baseline characteristics, intensive care measures and outcome parameters, differences between survivors and non-survivors from the intoxication group, as well as the individual groups of causative substances were investigated.ResultsIntoxicated patients were younger (median 41 vs. 66 years; p<0.001), more often female (55 vs. 45%; p<0.001), and normal weighted (36% vs. 30%; p<0.001), whereas more obese individuals where observed in the other group (37 vs. 31%; p<0.001). Intoxicated individuals had a significantly lower mortality compared to general ICU patients (1% vs. 10%; aOR 0.07 95%CI 0.05-0.11; p<0.001), a finding which persisted after multivariable adjustment (aOR 0.17 95%CI 0.12-0.24; p<0.001) and persisted in all subgroups. Markers of disease severity (SOFA-score: 3 (1-5) vs. 4 (2-6) pts.; p<0.001) and frequency of vasopressor use (5 vs. 15%; p<0.001) where lower, whereas rates of mechanical ventilation where higher (24 vs. 26%; p<0.001) in intoxicated individuals. There were no differences with regard to renal replacement therapy in the first three days (3 vs. 4%; p=0.26). In sensitivity analysis (interactions for age, sex, ethnicity, hospital category, maximum initial lactate, mechanical ventilation, and vasopressor use), a trend towards lower mortality in intoxicated patients persisted in all subgroups.ConclusionThis large-scale retrospective analysis indicates a significantly lower mortality of intoxicated individuals compared to general ICU patients.

A Machine Learning-Based Prediction Model for Acute Kidney Injury in Patients With Congestive Heart Failure.

BackgroundMachine learning (ML) has been used to build high performance prediction model. Patients with congestive heart failure (CHF) are vulnerable to acute kidney injury (AKI) which makes treatment difficult. We aimed to establish an ML-based prediction model for the early identification of AKI in patients with CHF.MethodsPatients data were extracted from the Medical Information Mart for Intensive Care III (MIMIC-III) database, and patients with CHF were selected. Comparisons between several common ML classifiers were conducted to select the best prediction model. Recursive feature elimination (RFE) was used to select important prediction features. The model was improved using hyperparameters optimization (HPO). The final model was validated using an external validation set from the eICU Collaborative Research Database. The area under the receiver operating characteristic curve (AUROC), accuracy, calibration curve and decision curve analysis were used to evaluate prediction performance. Additionally, the final model was used to predict renal replacement therapy (RRT) requirement and to assess the short-term prognosis of patients with CHF. Finally, a software program was developed based on the selected features, which could intuitively report the probability of AKI.ResultsA total of 8,580 patients with CHF were included, among whom 2,364 were diagnosed with AKI. The LightGBM model showed the best prediction performance (AUROC = 0.803) among the 13 ML-based models. After RFE and HPO, the final model was established with 18 features including serum creatinine (SCr), blood urea nitrogen (BUN) and urine output (UO). The prediction performance of LightGBM was better than that of measuring SCr, UO or SCr combined with UO (AUROCs: 0.809, 0.703, 0.560 and 0.714, respectively). Additionally, the final model could accurately predict RRT requirement in patients with (AUROC = 0.954). Moreover, the participants were divided into high- and low-risk groups for AKI, and the 90-day mortality in the high-risk group was significantly higher than that in the low-risk group (log-rank p < 0.001). Finally, external validation using the eICU database comprising 9,749 patients with CHF revealed satisfactory prediction outcomes (AUROC = 0.816).ConclusionA prediction model for AKI in patients with CHF was established based on LightGBM, and the prediction performance of this model was better than that of other models. This model may help in predicting RRT requirement and in identifying the population with poor prognosis among patients with CHF.

The association between the non-HDL-cholesterol to HDL-cholesterol ratio and 28-day mortality in sepsis patients: a cohort study

The focus of this study was to explore the association between the non-HDL-cholesterol to HDL-cholesterol (non-HDLc/HDLc) ratio and mortality in septic patients. This was a retrospective cohort study of patients with sepsis in the eICU Collaborative Research Database (eICU-CRD) from 208 distinct ICUs across the United States between 2014 and 2015 that explored. All-cause mortality within 28 days after ICU admission. A multivariable logistic regression model was used to estimate the risk of death. Of the 724 patients with a median age of 68 years, 43 (5.94%) died within 28 days after ICU admission. When the non-HDLc/HDLc ratio was < 3.3, the mortality rate decreased with an adjusted odds ratio (OR) of 0.60 (95% CI 0.37–0.99, P = 0.043) for every 1 increment in the non-HDLc/HDLc ratio. When the non-HDLc/HDLc ratio was ≥ 3.3, the mortality rate increased with an adjusted OR of 1.28 (95% CI 1.01–1.62, P = 0.039) for every one increment in the non-HDLc/HDLc ratio. For patients with sepsis, the association between the non-HDLc/HDLc ratio and the 28-day mortality risk was a U-shaped curve. A higher or lower non-HDLc/HDLc ratio was associated with an increased risk of 28-day mortality.

Systematic Review and Comparison of Publicly Available ICU Data Sets-A Decision Guide for Clinicians and Data Scientists.

As data science and artificial intelligence continue to rapidly gain traction, the publication of freely available ICU datasets has become invaluable to propel data-driven clinical research. In this guide for clinicians and researchers, we aim to: 1) systematically search and identify all publicly available adult clinical ICU datasets, 2) compare their characteristics, data quality, and richness and critically appraise their strengths and weaknesses, and 3) provide researchers with suggestions, which datasets are appropriate for answering their clinical question. A systematic search was performed in Pubmed, ArXiv, MedRxiv, and BioRxiv. We selected all studies that reported on publicly available adult patient-level intensive care datasets. A total of four publicly available, adult, critical care, patient-level databases were included (Amsterdam University Medical Center data base [AmsterdamUMCdb], eICU Collaborative Research Database eICU CRD], High time-resolution intensive care unit dataset [HiRID], and Medical Information Mart for Intensive Care-IV). Databases were compared using a priori defined categories, including demographics, patient characteristics, and data richness. The study protocol and search strategy were prospectively registered. Four ICU databases fulfilled all criteria for inclusion and were queried using SQL (PostgreSQL version 12; PostgreSQL Global Development Group) and analyzed using R (R Foundation for Statistical Computing, Vienna, Austria). The number of unique patient admissions varied between 23,106 (AmsterdamUMCdb) and 200,859 (eICU-CRD). Frequency of laboratory values and vital signs was highest in HiRID, for example, 5.2 (±3.4) lactate values per day and 29.7 (±10.2) systolic blood pressure values per hour. Treatment intensity varied with vasopressor and ventilatory support in 69.0% and 83.0% of patients in AmsterdamUMCdb versus 12.0% and 21.0% in eICU-CRD, respectively. ICU mortality ranged from 5.5% in eICU-CRD to 9.9% in AmsterdamUMCdb. We identified four publicly available adult clinical ICU datasets. Sample size, severity of illness, treatment intensity, and frequency of reported parameters differ markedly between the databases. This should guide clinicians and researchers which databases to best answer their clinical questions.

A recurrent machine learning model predicts intracranial hypertension in neurointensive care patients.

The evolution of intracranial pressure (ICP) of critically ill patients admitted to a neurointensive care unit (ICU) is difficult to predict. Besides the underlying disease and compromised intracranial space, ICP is affected by a multitude of factors, many of which are monitored on the ICU, but the complexity of the resulting patterns limits their clinical use. This paves the way for new machine learning techniques to assist clinical management of patients undergoing invasive ICP monitoring independent of the underlying disease.An institutional cohort (ICP-ICU) of patients with invasive ICP monitoring (n = 1346) was used to train recurrent machine learning models to predict the occurrence of ICP increases of ≥22 mmHg over a long (>2 h) time period in the upcoming hours. External validation was performed on patients undergoing invasive ICP measurement in two publicly available datasets [Medical Information Mart for Intensive Care (MIMIC, n = 998) and eICU Collaborative Research Database (n = 1634)].Different distances (1–24 h) between prediction time point and upcoming critical phase were evaluated, demonstrating a decrease in performance but still robust AUC-ROC with larger distances (24 h AUC-ROC: ICP-ICU 0.826 ± 0.0071, MIMIC 0.836 ± 0.0063, eICU 0.779 ± 0.0046, 1 h AUC-ROC: ICP-ICU 0.982 ± 0.0008, MIMIC 0.965 ± 0.0010, eICU 0.941 ± 0.0025). The model operates on sparse hourly data and is stable in handling variable input lengths and missingness through its nature of recurrence and internal memory. Calculation of gradient-based feature importance revealed individual underlying decisions for our long short time memory-based model and thereby provided improved clinical interpretability.Recurrent machine learning models have the potential to be an effective tool for the prediction of ICP increases with high translational potential.

A Simple Weaning Model Based on Interpretable Machine Learning Algorithm for Patients With Sepsis: A Research of MIMIC-IV and eICU Databases.

BackgroundInvasive mechanical ventilation plays an important role in the prognosis of patients with sepsis. However, there are, currently, no tools specifically designed to assess weaning from invasive mechanical ventilation in patients with sepsis. The aim of our study was to develop a practical model to predict weaning in patients with sepsis.MethodsWe extracted patient information from the Medical Information Mart for Intensive Care Database-IV (MIMIC-IV) and the eICU Collaborative Research Database (eICU-CRD). Kaplan–Meier curves were plotted to compare the 28-day mortality between patients who successfully weaned and those who failed to wean. Subsequently, MIMIC-IV was divided into a training set and an internal verification set, and the eICU-CRD was designated as the external verification set. We selected the best model to simplify the internal and external validation sets based on the performance of the model.ResultsA total of 5020 and 7081 sepsis patients with invasive mechanical ventilation in MIMIC-IV and eICU-CRD were included, respectively. After matching, weaning was independently associated with 28-day mortality and length of ICU stay (p < 0.001 and p = 0.002, respectively). After comparison, 35 clinical variables were extracted to build weaning models. XGBoost performed the best discrimination among the models in the internal and external validation sets (AUROC: 0.80 and 0.86, respectively). Finally, a simplified model was developed based on XGBoost, which included only four variables. The simplified model also had good predictive performance (AUROC:0.75 and 0.78 in internal and external validation sets, respectively) and was developed into a web-based tool for further review.ConclusionsWeaning success is independently related to short-term mortality in patients with sepsis. The simplified model based on the XGBoost algorithm provides good predictive performance and great clinical applicablity for weaning, and a web-based tool was developed for better clinical application.

Red Cell Distribution Width Is Independently Associated with Mortality in Sepsis

Background: Mortality in sepsis remains high. Studies on small cohorts have shown that red cell distribution width (RDW) is associated with mortality. The aim of this study was to validate these findings in a large multicenter cohort. Methods: We conducted this retrospective analysis of the multicenter eICU Collaborative Research Database in 16,423 septic patients. We split the cohort in patients with low (≤15%; n = 7,129) and high (>15%; n = 9,294) RDW. Univariable and multivariable multilevel logistic regressions were used to fit regression models for the binary primary outcome of hospital mortality and the secondary outcome intensive care unit (ICU) mortality with hospital unit as random effect. Optimal cutoffs were calculated using the Youden index. Results: Patients with high RDW were more often older than 65 years (57% vs. 50%; p < 0.001) and had higher Acute Physiology and Chronic Health Evaluation (APACHE) IV scores (69 vs. 60 pts.; p < 0.001). Both hospital (adjusted odds ratios [aOR] 1.18; 95% CI: 1.16–1.20; p < 0.001) and ICU mortality (aOR 1.16; 95% CI: 1.14–1.18; p < 0.001) were associated with RDW as a continuous variable. Patients with high RDW had a higher hospital mortality (20 vs. 9%; aOR 2.63; 95% CI: 2.38–2.90; p < 0.001). This finding persisted after multivariable adjustment (aOR 2.14; 95% CI: 1.93–2.37; p < 0.001) in a multilevel logistic regression analysis. The optimal RDW cutoff for the prediction of hospital mortality was 16%. Conclusion: We found an association of RDW with mortality in septic patients and propose an optimal cutoff value for risk stratification. In a combined model with lactate, RDW shows equivalent diagnostic performance to Sequential Organ Failure Assessment (SOFA) score and APACHE IV score.

A Machine Learning Approach for the Prediction of Traumatic Brain Injury Induced Coagulopathy.

Background: Traumatic brain injury-induced coagulopathy (TBI-IC), is a disease with poor prognosis and increased mortality rate.Objectives: Our study aimed to identify predictors as well as develop machine learning (ML) models to predict the risk of coagulopathy in this population.Methods: ML models were developed and validated based on two public databases named Medical Information Mart for Intensive Care (MIMIC)-IV and the eICU Collaborative Research Database (eICU-CRD). Candidate predictors, including demographics, family history, comorbidities, vital signs, laboratory findings, injury type, therapy strategy and scoring system were included. Models were compared on area under the curve (AUC), accuracy, sensitivity, specificity, positive and negative predictive values, and decision curve analysis (DCA) curve.Results: Of 999 patients in MIMIC-IV included in the final cohort, a total of 493 (49.35%) patients developed coagulopathy following TBI. Recursive feature elimination (RFE) selected 15 variables, including international normalized ratio (INR), prothrombin time (PT), sepsis related organ failure assessment (SOFA), activated partial thromboplastin time (APTT), platelet (PLT), hematocrit (HCT), red blood cell (RBC), hemoglobin (HGB), blood urea nitrogen (BUN), red blood cell volume distribution width (RDW), creatinine (CRE), congestive heart failure, myocardial infarction, sodium, and blood transfusion. The external validation in eICU-CRD demonstrated that adapting boosting (Ada) model had the highest AUC of 0.924 (95% CI: 0.902–0.943). Furthermore, in the DCA curve, the Ada model and the extreme Gradient Boosting (XGB) model had relatively higher net benefits (ie, the correct classification of coagulopathy considering a trade-off between false- negatives and false-positives)—over other models across a range of threshold probability values.Conclusions: The ML models, as indicated by our study, can be used to predict the incidence of TBI-IC in the intensive care unit (ICU).

Origin of Sepsis Associated with the Short-Term Mortality of Patients: A Retrospective Study Using the eICU Collaborative Research Database.

ObjectiveThe aim of this study was to compare the clinical characteristics and short-term mortality of patients with abdominal and pulmonary sepsis.DesignRetrospective cohort study.SettingAdult intensive care units (ICUs) at tertiary hospitals.ParticipantsAdult ICU patients from 2014 to 2015 in the eICU Collaborative Research Database.InterventionsIn univariate analysis, we compared the differences in the characteristics of patients in each group. Logistic regression models were used to evaluate the relationships between primary site of sepsis and short-term prognosis.Primary and Secondary Outcome MeasuresHospital and ICU survival.ResultsThe final dataset included 7023 pulmonary and 2360 abdominal sepsis patients, who accounted for 74.84% and 25.16%, respectively. We compared the results of the baseline characteristics, vital signs and laboratory indicators between the two groups. In the logistic regression models, we found that the hospital and ICU mortality of patients with abdominal sepsis was higher than that with pulmonary sepsis (p < 0.05, OR = 1.15, p < 0.05, OR = 1.19, respectively), although these results were no longer significantly after adjustment for confounders, but in the subgroups with SOFA score ≧ 8, the adjusted hospital mortality rate of patients with abdominal sepsis was 1.30 times higher than that of patients with pulmonary sepsis (p < 0.005, OR = 1.30, 95% CI 1.09–1.55), while there was no significant difference in the subgroups that SOFA score < 8.ConclusionIn the patients with SOFA score ≧ 8, the adjusted hospital mortality of patients with abdominal sepsis was higher than patients with pulmonary sepsis.

The Lactate/Albumin Ratio Predicts Mortality in Critically Ill Patients with Acute Kidney Injury: An Observational Multicenter Study on the eICU Database.

ObjectiveThe serum lactate/albumin ratio (LAR) can be used to independently predict mortality due to sepsis. However, whether the LAR predicts the outcomes of critically ill patients with acute kidney injury (AKI) remains unclear. This study was performed to assess the prognostic value of the LAR in critically ill AKI patients.MethodsThis retrospective observational study enrolled AKI patients, and all data were collected through the eICU Collaborative Research Database. Outcomes included in-hospital and intensive care unit (ICU) death. Multivariate Cox regression analysis was used to determine independent risk factors. Forest plots and smoothing curves were generated. A series of subgroup analyses were performed to further validate the robustness of the findings.ResultsA total of 4666 eligible patients were enrolled. We divided the participants into four groups according to the LAR: quartile (Q)1 (LAR < 0.46, n = 1167), Q2 (0.46 ≤ LAR < 0.79, n = 1162), Q3 (0.79 ≤ LAR < 1.49, n = 1170), and Q4 (LAR ≥ 1.49, n = 1167). The LAR, when analyzed as a continuous variable, was associated with hospital and ICU mortality (adjusted hazard ratio [HR] 1.22, 95% confidence interval [CI] 1.18–1.26, P < 0.0001 for both). The risk of in-hospital and ICU mortality increased with increasing LAR Q relative to Q1. The smoothing curves revealed a continuous linear association after adjusting for all covariates. By the Kaplan–Meier analysis, patients in the higher LAR group showed significantly shorter survival time. By the receiver operating characteristic analysis, LAR was efficient in predicting in-hospital mortality (area under the curve [AUC]: 0.717) and ICU mortality (AUC: 0.733). A positive and consistent effect of the LAR was seen in all subgroups analyses after adjusting for all covariates.ConclusionA high LAR is an independent risk factor for in-hospital and ICU mortality in critically ill patients with AKI. Further prospective studies are needed to validate these result.

Predicting Prolonged Length of ICU Stay through Machine Learning.

This study aimed to construct machine learning (ML) models for predicting prolonged length of stay (pLOS) in intensive care units (ICU) among general ICU patients. A multicenter database called eICU (Collaborative Research Database) was used for model derivation and internal validation, and the Medical Information Mart for Intensive Care (MIMIC) III database was used for external validation. We used four different ML methods (random forest, support vector machine, deep learning, and gradient boosting decision tree (GBDT)) to develop prediction models. The prediction performance of the four models were compared with the customized simplified acute physiology score (SAPS) II. The area under the receiver operation characteristic curve (AUROC), area under the precision-recall curve (AUPRC), estimated calibration index (ECI), and Brier score were used to measure performance. In internal validation, the GBDT model achieved the best overall performance (Brier score, 0.164), discrimination (AUROC, 0.742; AUPRC, 0.537), and calibration (ECI, 8.224). In external validation, the GBDT model also achieved the best overall performance (Brier score, 0.166), discrimination (AUROC, 0.747; AUPRC, 0.536), and calibration (ECI, 8.294). External validation showed that the calibration curve of the GBDT model was an optimal fit, and four ML models outperformed the customized SAPS II model. The GBDT-based pLOS-ICU prediction model had the best prediction performance among the five models on both internal and external datasets. Furthermore, it has the potential to assist ICU physicians to identify patients with pLOS-ICU risk and provide appropriate clinical interventions to improve patient outcomes.

Abstract 14410: Predicting Intubation and Ventilation Needs Upon Admission for Intensive Care Patients With Congestive Heart Failure

Introduction: Congestive heart failure (CHF) is one of the most common diagnoses for patients admitted to an intensive care unit (ICU), and intubation or ventilation may be necessary to combat severe associated respiratory symptoms. Current guidelines give high-level recommendations to intubate patients with a Glasgow Coma Scale (GCS) score of 8 or lower. However, prior studies have shown that a large portion of patients who were ultimately intubated had GCS scores greater than 8, demonstrating the need for a better tool to predict the necessity of intubation. The present study sought to build machine-learning (ML) models that could predict the need for intubation or ventilation for CHF patients upon ICU admission. Methods: 5,623 patients (of which 1,058 were intubated and 2,816 were ventilated) in the eICU Collaborative Research Database who underwent screening using the Acute Physiology and Chronic Health Evaluation and were diagnosed with CHF were used to train (50%) and test (50%) the ML models. Gender, age, treatment status, readmission status, medication status upon admission, GCS motor score upon admission, and GCS eyes score upon admission were used as feature inputs for the ML models, which were classic three-layer neural networks. For intubation, the ML model performance was compared against using the common heuristic of intubating patients with a GCS score of 8 or less. Results: The ML model for intubation upon admission prediction had an accuracy of 84.8% and an area under the receiver operating characteristic curve (AUC) of 0.79. Meanwhile, using the aforementioned heuristic for intubation prediction yields an accuracy of 85.7% (AUC n/a). A second ML model was trained to predict the need for ventilation upon admission and had an accuracy of 71.2% and an AUC of 0.80. Conclusion: This newly developed ML model for intubation demonstrates relative success in predicting intubation needs for patients with CHF upon ICU admission. In addition, this model performs similarly to currently accepted guidelines. Further improvement and a larger dataset could better train the model to provide more accurate predictions. Such predictions could help support medical decision-making and resource allocation planning.

Mortality Prediction Using SaO2/FiO2 Ratio Based on eICU Database Analysis.

Purpose PaO2 to FiO2 ratio (P/F) is used to assess the degree of hypoxemia adjusted for oxygen requirements. The Berlin definition of Acute Respiratory Distress Syndrome (ARDS) includes P/F as a diagnostic criterion. P/F is invasive and cost-prohibitive for resource-limited settings. SaO2/FiO2 (S/F) ratio has the advantages of being easy to calculate, noninvasive, continuous, cost-effective, and reliable, as well as lower infection exposure potential for staff, and avoids iatrogenic anemia. Previous work suggests that the SaO2/FiO2 ratio (S/F) correlates with P/F and can be used as a surrogate in ARDS. Quantitative correlation between S/F and P/F has been verified, but the data for the relative predictive ability for ICU mortality remains in question. We hypothesize that S/F is noninferior to P/F as a predictive feature for ICU mortality. Using a machine-learning approach, we hope to demonstrate the relative mortality predictive capacities of S/F and P/F. Methods We extracted data from the eICU Collaborative Research Database. The features age, gender, SaO2, PaO2, FIO2, admission diagnosis, Apache IV, mechanical ventilation (MV), and ICU mortality were extracted. Mortality was the dependent variable for our prediction models. Exploratory data analysis was performed in Python. Missing data was imputed with Sklearn Iterative Imputer. Random assignment of all the encounters, 80% to the training (n = 26690) and 20% to testing (n = 6741), was stratified by positive and negative classes to ensure a balanced distribution. We scaled the data using the Sklearn Standard Scaler. Categorical values were encoded using Target Encoding. We used a gradient boosting decision tree algorithm variant called XGBoost as our model. Model hyperparameters were tuned using the Sklearn RandomizedSearchCV with tenfold cross-validation. We used AUC as our metric for model performance. Feature importance was assessed using SHAP, ELI5 (permutation importance), and a built-in XGBoost feature importance method. We constructed partial dependence plots to illustrate the relationship between mortality probability and S/F values. Results The XGBoost hyperparameter optimized model had an AUC score of .85 on the test set. The hyperparameters selected to train the final models were as follows: colsample_bytree of 0.8, gamma of 1, max_depth of 3, subsample of 1, min_child_weight of 10, and scale_pos_weight of 3. The SHAP, ELI5, and XGBoost feature importance analysis demonstrates that the S/F ratio ranks as the strongest predictor for mortality amongst the physiologic variables. The partial dependence plots illustrate that mortality rises significantly above S/F values of 200. Conclusion S/F was a stronger predictor of mortality than P/F based upon feature importance evaluation of our data. Our study is hypothesis-generating and a prospective evaluation is warranted. Take-Home Points. S/F ratio is a noninvasive continuous method of measuring hypoxemia as compared to P/F ratio. Our study shows that the S/F ratio is a better predictor of mortality than the more widely used P/F ratio to monitor and manage hypoxemia.

Local Differential Privacy in the Medical Domain to Protect Sensitive Information: Algorithm Development and Real-World Validation.

BackgroundPrivacy is of increasing interest in the present big data era, particularly the privacy of medical data. Specifically, differential privacy has emerged as the standard method for preservation of privacy during data analysis and publishing.ObjectiveUsing machine learning techniques, we applied differential privacy to medical data with diverse parameters and checked the feasibility of our algorithms with synthetic data as well as the balance between data privacy and utility.MethodsAll data were normalized to a range between –1 and 1, and the bounded Laplacian method was applied to prevent the generation of out-of-bound values after applying the differential privacy algorithm. To preserve the cardinality of the categorical variables, we performed postprocessing via discretization. The algorithm was evaluated using both synthetic and real-world data (from the eICU Collaborative Research Database). We evaluated the difference between the original data and the perturbated data using misclassification rates and the mean squared error for categorical data and continuous data, respectively. Further, we compared the performance of classification models that predict in-hospital mortality using real-world data.ResultsThe misclassification rate of categorical variables ranged between 0.49 and 0.85 when the value of ε was 0.1, and it converged to 0 as ε increased. When ε was between 102 and 103, the misclassification rate rapidly dropped to 0. Similarly, the mean squared error of the continuous variables decreased as ε increased. The performance of the model developed from perturbed data converged to that of the model developed from original data as ε increased. In particular, the accuracy of a random forest model developed from the original data was 0.801, and this value ranged from 0.757 to 0.81 when ε was 10-1 and 104, respectively.ConclusionsWe applied local differential privacy to medical domain data, which are diverse and high dimensional. Higher noise may offer enhanced privacy, but it simultaneously hinders utility. We should choose an appropriate degree of noise for data perturbation to balance privacy and utility depending on specific situations.

Influence of fluid balance on the prognosis of patients with sepsis

BackgroundEarly and timely fluid treatment or resuscitation are the basic measures for the active treatment of sepsis. Our aim is to further explore the relationship between fluid balance and prognosis in patients with sepsis on a daily basis for 5 days.MethodsSepsis patients in eICU Collaborative Research Database were divided into the negative balance group (NB/−) and the positive balance group (PB/+) according to daily fluid balance. The primary outcome was in-hospital mortality. Survival differences between the groups were analyzed by using Cox regression. Then dose-response relationship between fluid balance and in-hospital mortality was studied using restricted cubic splines (RCSs). Furthermore, patients with fluid balance data for the previous three consecutive days were selected and divided into eight groups (“+/+/+”, “+/+/−”, “+/ −/−”, “+/ −/+”, “−/ −/−”, “−/ −/+”, “−/+/+”, and “−/+/−”). Kaplan–Meier curves and Cox regression were used to show the survival difference between groups.ResultsOur study, which included 19,557 patients in a multicenter database, showed that positive fluid balances on days 1, 2, and 3 after sepsis diagnosis were associated with poor prognosis with the HRs of 1.29 (1.20,1.40), 1.13 (1.01,1.27), and 1.25 (1.08,1.44), respectively, while the fluid balance on days 4 and 5 had no effect on the primary outcome. Then RCSs showed an overall trend that the risk of in-hospital mortality on days 1, 2, and 3 increased with increasing fluid balance. For three consecutive days of fluid balance, we studied 9205 patients and Kaplan–Meier curves revealed survival differences among patients in the eight groups. The cox model demonstrated that compared with the “+/+/+” group, the “+/ −/−”, “−/ −/−”, “−/ −/+”, “−/+/+”, and “−/+/−” groups had a lower risk of in-hospital mortality, with HRs of 0.65 (0.45,0.93), 0.72 (0.60,0.86), 0.63 (0.43,0.93), 0.69 (0.48,0.98), and 0.63 (0.42,0.96), respectively.ConclusionsIn patients with sepsis, positive fluid balance on days 1, 2, and 3 was associated with adverse outcomes. For patients with fluid balance for three consecutive days, the “+/−/−”, “−/ −/−”, “−/−/+”, “−/+/+”, and “−/+/−” groups were less likely to die in hospital than the “+/+/+” group.