Predictors of Mortality Using Machine Learning Decision Tree Algorithm in Critically Ill Adult Patients with COVID-19 Admitted to the ICU.

Background:The Coronavirus Disease-19 (COVID-19) caused by the Severe Acute Respiratory Syndrome Coronavirus 2 (SARS-CoV-2) is a major cause of intensive care unit (ICU) admissions globally. Robust data of epidemiology, characteristics, and disease outcomes from different regions and populations showed considerable variations. However, limited number of reports addressed predictors of mortality utilizing machine learning methods. Herein, we aimed to describe the association and relationship of a predefined set of variables found to be predictive of 28–day ICU outcome among adults COVID-19 patients admitted to the ICU using a machine learning decision tree (DT) algorithm.Methods:This was a prospective/retrospective, multicenter cohort study from 14 hospitals in Saudi Arabia. We included critically ill COVID-19 patients admitted to the ICU between March 1, 2020, and October 31, 2020. The primary outcome was 28-day ICU mortality. Secondary outcomes were 90-day mortality and ICU length of stay. The predictors of mortality were identified using two predictive models, the conventional logistic regression and DT analysis.Results:A total of 1468 critically ill COVID-19 patients were included. The mean age was 55.9 (SD±15.1) years, with 74% of the patients were males. The 28-day ICU mortality was 540 (36.8%), while 90-day mortality was 600 (40.9%). The multivariable logistic regression model demonstrated that the PaO2/FiO2 ratio on ICU admission and the need for intubation or vasopressors could strongly predict 28-day ICU mortality. The DT algorithm identified five variables [need for intubation, need for vasopressors, age, gender, and PaO2/FiO2 ratio] provided in an algorithmic fashion to predict 28-day ICU outcome. Conclusion:Five clinical predictors of 28-day ICU outcome were identified using DT algorithmic analysis of COVID-19 patients admitted to ICU. The findings of this DT analysis may be used in ICU for early identification of critically ill COVID-19 patients who are at high risk of 28-day mortality.

Machine learning identifies ICU outcome predictors in a multicenter COVID-19 cohort

Background Intensive Care Resources are heavily utilized during the COVID-19 pandemic. However, risk stratification and prediction of SARS-CoV-2 patient clinical outcomes upon ICU admission remain inadequate. This study aimed to develop a machine learning model, based on retrospective & prospective clinical data, to stratify patient risk and predict ICU survival and outcomes. Methods A Germany-wide electronic registry was established to pseudonymously collect admission, therapeutic and discharge information of SARS-CoV-2 ICU patients retrospectively and prospectively. Machine learning approaches were evaluated for the accuracy and interpretability of predictions. The Explainable Boosting Machine approach was selected as the most suitable method. Individual, non-linear shape functions for predictive parameters and parameter interactions are reported. Results 1039 patients were included in the Explainable Boosting Machine model, 596 patients retrospectively collected, and 443 patients prospectively collected. The model for prediction of general ICU outcome was shown to be more reliable to predict “survival”. Age, inflammatory and thrombotic activity, and severity of ARDS at ICU admission were shown to be predictive of ICU survival. Patients’ age, pulmonary dysfunction and transfer from an external institution were predictors for ECMO therapy. The interaction of patient age with D-dimer levels on admission and creatinine levels with SOFA score without GCS were predictors for renal replacement therapy. Conclusions Using Explainable Boosting Machine analysis, we confirmed and weighed previously reported and identified novel predictors for outcome in critically ill COVID-19 patients. Using this strategy, predictive modeling of COVID-19 ICU patient outcomes can be performed overcoming the limitations of linear regression models. Trial registration “ClinicalTrials” (clinicaltrials.gov) under NCT04455451.

Association of Troponin T levels and functional outcome 3 months after subarachnoid hemorrhage

TroponinT levels are frequently elevated after subarachnoid hemorrhage (SAH). However, their clinical impact on long term outcomes still remains unclear. This study evaluates the association of TroponinT and functional outcomes 3 months after SAH. Data were obtained in the frame of a randomized controlled trial exploring the association of Goal-directed hemodynamic therapy and outcomes after SAH (NCT01832389). TroponinT was measured daily for the first 14 days after admission or until discharge from the ICU. Outcome was assessed using Glasgow Outcome Scale (GOS) 3 months after discharge. Logistic regression was used to explore the association between initial TroponinT values stratified by tertiles and admission as well as outcome parameters. TroponinT measurements were analyzed in 105 patients. TroponinT values at admission were associated with outcome assessed by GOS in a univariate analysis. TroponinT was not predictive of vasospasm or delayed cerebral ischemia, but an association with pulmonary and cardiac complications was observed. After adjustment for age, history of arterial hypertension and World Federation of Neurosurgical Societies (WFNS) grade, TroponinT levels at admission were not independently associated with worse outcome (GOS 1–3) or death at 3 months. In summary, TroponinT levels at admission are associated with 3 months-GOS but have limited ability to independently predict outcome after SAH.

Machine Learning Identifies ICU Outcome Predictors in a Multicenter COVID-19 Cohort.

Background. Intensive Care Resources are heavily utilized during the COVID-19 pandemic. However, risk stratification and prediction of SARS-CoV-2 patient clinical outcomes upon ICU admission remain inadequate. This study aimed to develop a machine learning model, based on retrospective & prospective clinical data, to stratify patient risk and predict ICU survival and outcomes.Methods. A Germany-wide electronic registry was established to pseudonymously collect admission, therapeutic and discharge information of SARS-CoV-2 ICU patients retrospectively and prospectively. Machine learning approaches were evaluated for the accuracy and interpretability of predictions. The Explainable Boosting Machine approach was selected as the most suitable method. Individual, non-linear shape functions for predictive parameters and parameter interactions are reported. Results. 1,039 patients were included in the Explainable Boosting Machine model, 596 patients retrospectively collected, and 443 patients prospectively collected. The model for prediction of general ICU outcome was shown to be more reliable to predict “survival”. Age, inflammatory and thrombotic activity, and severity of ARDS at ICU admission were shown to be predictive of ICU survival. Patients’ age, pulmonary dysfunction and transfer from an external institution were predictors for ECMO therapy. The interaction of patient age with D-dimer levels on admission and creatinine levels with SOFA score without GCS were predictors for renal replacement therapy. Conclusions. Using Explainable Boosting Machine analysis, we confirmed and weighed previously reported and identified novel predictors for outcome in critically ill COVID-19 patients. Using this strategy, predictive modeling of COVID-19 ICU patient outcomes can be performed overcoming the limitations of linear regression models.Trial registration. “ClinicalTrials” (clinicaltrials.gov) under NCT04455451

Mechanical Power during Veno-Venous Extracorporeal Membrane Oxygenation Initiation: A Pilot-Study

Mechanical power (MP) represents a useful parameter to describe and quantify the forces applied to the lungs during mechanical ventilation (MV). In this multi-center, prospective, observational study, we analyzed MP variations following MV adjustments after veno-venous extra-corporeal membrane oxygenation (VV ECMO) initiation. We also investigated whether the MV parameters (including MP) in the early phases of VV ECMO run may be related to the intensive care unit (ICU) mortality. Thirty-five patients with severe acute respiratory distress syndrome were prospectively enrolled and analyzed. After VV ECMO initiation, we observed a significant decrease in median MP (32.4 vs. 8.2 J/min, p < 0.001), plateau pressure (27 vs. 21 cmH2O, p = 0.012), driving pressure (11 vs. 8 cmH2O, p = 0.014), respiratory rate (RR, 22 vs. 14 breaths/min, p < 0.001), and tidal volume adjusted to patient ideal body weight (VT/IBW, 5.5 vs. 4.0 mL/kg, p = 0.001) values. During the early phase of ECMO run, RR (17 vs. 13 breaths/min, p = 0.003) was significantly higher, while positive end-expiratory pressure (10 vs. 14 cmH2O, p = 0.048) and VT/IBW (3.0 vs. 4.0 mL/kg, p = 0.028) were lower in ICU non-survivors, when compared to the survivors. The observed decrease in MP after ECMO initiation did not influence ICU outcome. Waiting for large studies assessing the role of these parameters in VV ECMO patients, RR and MP monitoring should not be underrated during ECMO.

Biomarkers of Neurological Outcome After Aneurysmal Subarachnoid Hemorrhage as Early Predictors at Discharge from an Intensive Care Unit

Background Subarachnoid bleeding is associated with brain injuries and ranges from almost negligible to acute and life threatening. The main objectives were to study changes in brain-specific biomarker levels in patients after an aneurysmal subarachnoid hemorrhage (aSAH) in relation to early clinical findings, severity scores, and intensive care unit (ICU) outcome. Analysis was done to identify specific biomarkers as predictors of a bad outcome in the acute treatment phase. Methods Analysis was performed for the proteins of neurofilament, neuron-specific enolase (NSE), microtubule-associated protein tau (MAPT), and for the proteins of glial cells, S100B, and glial fibrillary acidic protein (GFAP). Outcomes were assessed at discharge from the ICU and analyzed based on the grade in the Glasgow Outcome Scale (GOS). Patients were classified into two groups: with a good outcome (Group 1: GOS IV–V, n = 24) and with a bad outcome (Group 2: GOS I–III, n = 31). Blood samples were taken upon admission to the ICU and afterward daily for up to 6 days. Results In Group 1, the level of S100B (1.0, 0.9, 0.7, 2.0, 1.0, 0.3 ng/mL) and NSE (1.5, 2.0, 1.6, 1.2, 16.6, 2.2 ng/mL) was significantly lower than in Group 2 (S100B: 4.7, 4.8, 4.4, 4.5, 6.6, 6.8 ng/mL; NSE: 4.0, 4.1, 4.3, 3.8, 4.4, 2.5 1.1 ng/mL) on day 1–6, respectively. MAPT was significantly lower only on the first and second day (83.2 ± 25.1, 132.7 ± 88.1 pg/mL in Group 1 vs. 625.0 ± 250.7, 616.4 ± 391.6 pg/mL in Group 2). GFAP was elevated in both groups from day 1 to 6. In the ROC analysis, S100B showed the highest ability to predict bad ICU outcome of the four biomarkers measured on admission [area under the curve (AUC) 0.81; 95% CI 0.67–0.94, p < 0.001]. NSE and MAPT also had significant predictive value (AUC 0.71; 95% CI 0.54–0.87, p = 0.01; AUC 0.74; 95% CI 0.55–0.92, p = 0.01, respectively). A strong negative correlation between the GOS and S100B and the GOS and NSE was recorded on days 1–5, and between the GOS and MAPT on day 1. Conclusion Our findings provide evidence that brain biomarkers such as S100B, NSE, GFAP, and MAPT increase significantly in patients following aSAH. There is a direct relationship between the neurological outcome in the acute treatment phase and the levels of S100B, NSE, and MAPT. The detection of brain-specific biomarkers in conjunction with clinical data may constitute a valuable diagnostic and prognostic tool in the early phase of aSAH treatment.

Between-centre differences for COVID-19 ICU mortality from early data in England

The high numbers of COVID-19 patients developing severe respiratory failure has placed exceptional demands on ICU capacity around the world. Understanding the determinants of ICU mortality is important for surge planning and shared decision making. We used early data from the COVID-19 Hospitalisation in England Surveillance System (from the start of data collection 8th February -22nd May 2020) to look for factors associated with ICU outcome in the hope that information from such timely analysis may be actionable before the outbreak peak. Immunosuppressive disease, chronic cardiorespiratory/renal disease and age were key determinants of ICU mortality in a proportional hazards mixed effects model. However variation in site-stratified random effects were comparable in magnitude suggesting substantial between-centre variability in mortality. Notwithstanding possible ascertainment and lead-time effects, these early results motivate comparative effectiveness research to understand the origin of such differences and optimise surge ICU provision.