Dataset Information

Mortality prediction model for the triage of COVID-19, pneumonia, and mechanically ventilated ICU patients: A retrospective study.

ABSTRACT:

Rationale

Prediction of patients at risk for mortality can help triage patients and assist in resource allocation.

Objectives

Develop and evaluate a machine learning-based algorithm which accurately predicts mortality in COVID-19, pneumonia, and mechanically ventilated patients.

Methods

Retrospective study of 53,001 total ICU patients, including 9166 patients with pneumonia and 25,895 mechanically ventilated patients, performed on the MIMIC dataset. An additional retrospective analysis was performed on a community hospital dataset containing 114 patients positive for SARS-COV-2 by PCR test. The outcome of interest was in-hospital patient mortality.

Results

When trained and tested on the MIMIC dataset, the XGBoost predictor obtained area under the receiver operating characteristic (AUROC) values of 0.82, 0.81, 0.77, and 0.75 for mortality prediction on mechanically ventilated patients at 12-, 24-, 48-, and 72- hour windows, respectively, and AUROCs of 0.87, 0.78, 0.77, and 0.734 for mortality prediction on pneumonia patients at 12-, 24-, 48-, and 72- hour windows, respectively. The predictor outperformed the qSOFA, MEWS and CURB-65 risk scores at all prediction windows. When tested on the community hospital dataset, the predictor obtained AUROCs of 0.91, 0.90, 0.86, and 0.87 for mortality prediction on COVID-19 patients at 12-, 24-, 48-, and 72- hour windows, respectively, outperforming the qSOFA, MEWS and CURB-65 risk scores at all prediction windows.

Conclusions

This machine learning-based algorithm is a useful predictive tool for anticipating patient mortality at clinically useful timepoints, and is capable of accurate mortality prediction for mechanically ventilated patients as well as those diagnosed with pneumonia and COVID-19.

SUBMITTER: Ryan L

PROVIDER: S-EPMC7532803 | biostudies-literature |

REPOSITORIES: biostudies-literature

ACCESS DATA

Similar Datasets

Project description:PurposeWhile fever may be a presenting symptom of COVID-19, fever at hospital admission has not been identified as a predictor of mortality. However, hyperthermia during critical illness among ventilated COVID-19 patients in the ICU has not yet been studied. We sought to determine mortality predictors among ventilated COVID-19 ICU patients and we hypothesized that fever in the ICU is predictive of mortality.Materials and methodsWe conducted a retrospective cohort study of 103 ventilated COVID-19 patients admitted to the ICU between March 14 and May 27, 2020. Final follow-up was June 5, 2020. Patients discharged from the ICU or who died were included. Patients still admitted to the ICU at final follow-up were excluded.Results103 patients were included, 40 survived and 63(61.1%) died. Deceased patients were older {66 years[IQR18] vs 62.5[IQR10], (p = 0.0237)}, more often male {48(68%) vs 22(55%), (p = 0.0247)}, had lower initial oxygen saturation {86.0%[IQR18] vs 91.5%[IQR11.5], (p = 0.0060)}, and had lower pH nadir than survivors {7.10[IQR0.2] vs 7.30[IQR0.2] (p < 0.0001)}. Patients had higher peak temperatures during ICU stay as compared to hospital presentation {103.3°F[IQR1.7] vs 100.0°F[IQR3.5], (p < 0.0001)}. Deceased patients had higher peak ICU temperatures than survivors {103.6°F[IQR2.0] vs 102.9°F[IQR1.4], (p = 0.0008)}. Increasing peak temperatures were linearly associated with mortality. Febrile patients who underwent targeted temperature management to achieve normothermia did not have different outcomes than those not actively cooled. Multivariable analysis revealed 60% and 75% higher risk of mortality with peak temperature greater than 103°F and 104°F respectively; it also confirmed hyperthermia, age, male sex, and acidosis to be predictors of mortality.ConclusionsThis is one of the first studies to identify ICU hyperthermia as predictive of mortality in ventilated COVID-19 patients. Additional predictors included male sex, age, and acidosis. With COVID-19 cases increasing, identification of ICU mortality predictors is crucial to improve risk stratification, resource management, and patient outcomes.

Project description:BackgroundMechanical ventilation (MV) is vital for critically ill ICU patients but carries significant mortality risks. This study aims to develop a predictive model to estimate hospital mortality among MV patients, utilizing comprehensive health data to assist ICU physicians with early-stage alerts.MethodsWe developed a Machine Learning (ML) framework to predict hospital mortality in ICU patients receiving MV. Using the MIMIC-III database, we identified 25,202 eligible patients through ICD-9 codes. We employed backward elimination and the Lasso method, selecting 32 features based on clinical insights and literature. Data preprocessing included eliminating columns with over 90% missing data and using mean imputation for the remaining missing values. To address class imbalance, we used the Synthetic Minority Over-sampling Technique (SMOTE). We evaluated several ML models, including CatBoost, XGBoost, Decision Tree, Random Forest, Support Vector Machine (SVM), K-Nearest Neighbors (KNN), and Logistic Regression, using a 70/30 train-test split. The CatBoost model was chosen for its superior performance in terms of accuracy, precision, recall, F1-score, AUROC metrics, and calibration plots.ResultsThe study involved a cohort of 25,202 patients on MV. The CatBoost model attained an AUROC of 0.862, an increase from an initial AUROC of 0.821, which was the best reported in the literature. It also demonstrated an accuracy of 0.789, an F1-score of 0.747, and better calibration, outperforming other models. These improvements are due to systematic feature selection and the robust gradient boosting architecture of CatBoost.ConclusionThe preprocessing methodology significantly reduced the number of relevant features, simplifying computational processes, and identified critical features previously overlooked. Integrating these features and tuning the parameters, our model demonstrated strong generalization to unseen data. This highlights the potential of ML as a crucial tool in ICUs, enhancing resource allocation and providing more personalized interventions for MV patients.

Project description:BackgroundBiomarkers can be used to detect the presence of endothelial and/or alveolar epithelial injuries in case of ARDS. Angiopoietin-2 (Ang-2), soluble intercellular adhesion molecule-1 (ICAM-1), vascular cell adhesion protein-1 (VCAM-1), P-selectin and E-selectin are biomarkers of endothelial injury, whereas the receptor for advanced glycation end-products (RAGE) reflects alveolar epithelial injury. The aims of this study were to evaluate whether the plasma concentration of the above-mentioned biomarkers was different 1) in survivors and non-survivors of COVID-19-related ARDS and 2) in COVID-19-related and classical ARDS.MethodsThis prospective study was performed in two COVID-19-dedicated Intensive Care Units (ICU) and one non-COVID-19 ICU at Ferrara University Hospital. A cohort of 31 mechanically ventilated patients with COVID-19 ARDS and a cohort of 11 patients with classical ARDS were enrolled. Ang-2, ICAM-1, VCAM-1, P-selectin, E-selectin and RAGE were determined with a bead-based multiplex immunoassay at three time points: inclusion in the study (T1), after 7 ± 2 days (T2) and 14 ± 2 days (T3). The primary outcome was to evaluate the plasma trend of the biomarker levels in survivors and non-survivors. The secondary outcome was to evaluate the differences in respiratory mechanics variables and gas exchanges between survivors and non-survivors. Furthermore, we compared the plasma levels of the biomarkers at T1 in patients with COVID-19-related ARDS and classical ARDS.ResultsIn COVID-19-related ARDS, the plasma levels of Ang-2 and ICAM-1 at T1 were statistically higher in non-survivors than survivors, (p = 0.04 and p = 0.03, respectively), whereas those of P-selectin, E-selectin and RAGE did not differ. Ang-2 and ICAM-1 at T1 were predictors of mortality (AUROC 0.650 and 0.717, respectively). At T1, RAGE and P-selectin levels were higher in classical ARDS than in COVID-19-related ARDS. Ang-2, ICAM-1 and E-selectin were lower in classical ARDS than in COVID-19-related ARDS (all p < 0.001).ConclusionsCOVID-19 ARDS is characterized by an early pulmonary endothelial injury, as detected by Ang-2 and ICAM-1. COVID-19 ARDS and classical ARDS exhibited a different expression of biomarkers, suggesting different pathological pathways. Trial registration NCT04343053 , Date of registration: April 13, 2020.

Project description:ObjectivesAssess the association between tocilizumab administration and clinical outcomes among mechanically ventilated patients with COVID-19 pneumonia.DesignRetrospective cohort study.SettingLarge integrated health system with 9 million members in California, USA.Participants4185 Kaiser Permanente members hospitalised with COVID-19 pneumonia requiring invasive mechanical ventilation (IMV).InterventionsReceipt of tocilizumab within 10 days of initiation of IMV.Outcome measuresUsing a retrospective cohort of consecutive patients hospitalised with COVID-19 pneumonia who required IMV in a large integrated health system in California, USA, we assessed the association between tocilizumab administration and 28-day mortality, time to extubation from IMV and time to hospital discharge.ResultsAmong 4185 patients, 184 received tocilizumab and 4001 patients did not receive tocilizumab within 10 days of initiation of IMV. After inverse probability weighting, baseline characteristics were well balanced between groups. Patients treated with tocilizumab had a similar risk of death in the 28 days after intubation compared with patients not treated with tocilizumab (adjusted HR (aHR), 1.21, 95% CI 0.98 to 1.50), but did have a significantly longer time-to-extubation (aHR 0.71; 95% CI 0.57 to 0.88) and time-to-hospital-discharge (aHR 0.66; 95% CI 0.50 to 0.88). However, patients treated with tocilizumab ≤2 days after initiation of IMV had a similar risk of mortality (aHR 1.47; 95% CI 0.96 to 2.26), but significantly shorter time-to-extubation (aHR 0.37; 95% CI 0.23 to 0.58) and time-to-hospital-discharge (aHR 0.31; 95% CI CI 0.17 to 0.56) compared with patients treated with tocilizumab 3-10 days after initiation of IMV.ConclusionsAmong mechanically ventilated patients with COVID-19, the risk of death in the 28-day follow-up period was similar, but time-to-extubation and time-to-hospital-discharge were longer in patients who received tocilizumab within 10 days of initiation of IMV compared with patients who did not receive tocilizumab.

Project description:BackgroundIn most countries, patients receiving mechanical ventilation (MV) are treated in intensive care units (ICUs). However, in some countries, including Japan, many patients on MV are not treated in ICUs. There are insufficient epidemiological data on these patients. Here, we sought to describe the epidemiology of patients on MV in Japan by comparing and contrasting patients on MV treated in ICUs and in non-ICU settings. A preliminary comparison of patient outcomes between ICU and non-ICU patients was a secondary objective.MethodsData on adult patients receiving MV for at least 3 days in ICUs or non-ICU settings from April 2010 through March 2012 were obtained from the Quality Indicator/Improvement Project, a voluntary data-administration project covering more than 400 acute-care hospitals in Japan. We excluded patients with cancer-related diagnoses. Patient demographic data and the critical care provided were compared between groups.ResultsOver the study period, 17,775 patients on MV were treated only in non-ICU settings, whereas 20,516 patients were treated at least once in ICUs (46.4% vs. 53.6%). Average age was higher in non-ICU patients than in ICU patients (72.8 vs. 70.2, P?<?0.001). Mean number of ventilation days was greater in non-ICU patients (11.7 vs. 9.5, P?<?0.001). Hospital mortality was higher in non-ICU patients (41.4% vs. 38.8%, P?<?0.001). Standard critical care (e.g., arterial line placement, enteral nutrition, and stress-ulcer prevention) was provided significantly less often in non-ICU patients. Multivariate analysis showed that ICU admission significantly decreased hospital mortality (adjusted odds ratio 0.713, 95% CI 0.676 to 0.753).ConclusionsA large proportion of Japanese patients on MV were treated in non-ICU settings. Analysis of administrative data indicated preliminarily that hospital mortality rates in these patients were higher in non-ICU settings than in ICUs. Prospective analyses comparing non-ICU and ICU patients on MV by severity scoring are needed.