Article

Disease-course adapting machine learning prognostication models in critically ill elderly COVID-19 patients: a multi-centre cohort study with external validation

Authors:
To read the full-text of this research, you can request a copy directly from the authors.

Abstract

Background: The SARS-CoV-2 coronavirus disease (COVID-19) pandemic is challenging health care systems globally. The disease disproportionately affects the elderly population, both in terms of disease severity and mortality risk. Objective: This study aimed to evaluate machine-learning based prognostication models for critically ill elderly COVID-19 patients, which dynamically incorporated multifaceted clinical information on the evolution of the disease. Methods: This multi-centre cohort study obtained patient data from 151 ICUs from 26 countries (COVIP study). Different models based on the Sequential Organ Failure Assessment (SOFA), Logistic Regression (LR), Random Forest (RF) and Extreme Gradient Boosting (XGBoost) were derived as baseline models that included admission variables only. We subsequently included clinical events and time-to-event as additional variables to derive the final models using the same algorithms and compared their performance with the baseline group. Furthermore, we derived baseline and final models on a European patient cohort and externally validated them on a non-European cohort that included Asian, African and American patients. Results: In total, 1,432 elderly (≥70 years) COVID-19 positive patients were admitted to an intensive care unit. Of these 809 (56.5%) patients survived up to 30 days after admission. The average length of stay was 21.6 (±18.2) days. Final models that incorporated clinical events and time-to-event provided superior performance with AUC of 0.81 (95% CI 0.804-0.811), with respect to both, the baseline models that used admission variables only and conventional ICU prediction models (SOFA-score, p<.001). The average precision increased from 0.65 (95% CI 0.650-0.655) to 0.77 (95% CI 0.759-0.770). Conclusions: Integrating important clinical events and time-to-event information led to a superior accuracy of 30-day mortality prediction compared with models based on the admission information and conventional ICU prediction models. The present study shows that machine-learning models provide additional information and may support complex decision-making in critically ill elderly COVID-19 patients. Clinicaltrial: Nct04321265.

No full-text available

Request Full-text Paper PDF

To read the full-text of this research,
you can request a copy directly from the authors.

ResearchGate has not been able to resolve any citations for this publication.
Article
Full-text available
Background: The primary aim of this study was to assess the outcome of elderly intensive care unit (ICU) patients treated during the spring and autumn COVID-19 surges in Europe. Methods: This was a prospective European observational study (the COVIP study) in ICU patients aged 70 years and older admitted with COVID-19 disease from March to December 2020 to 159 ICUs in 14 European countries. An electronic database was used to register a number of parameters including: SOFA score, Clinical Frailty Scale, co-morbidities, usual ICU procedures and survival at 90 days. The study was registered at ClinicalTrials.gov (NCT04321265). Results: In total, 2625 patients were included, 1327 from the first and 1298 from the second surge. Median age was 74 and 75 years in surge 1 and 2, respectively. SOFA score was higher in the first surge (median 6 versus 5, p < 0.0001). The PaO2/FiO2 ratio at admission was higher during surge 1, and more patients received invasive mechanical ventilation (78% versus 68%, p < 0.0001). During the first 15 days of treatment, survival was similar during the first and the second surge. Survival was lower in the second surge after day 15 and differed after 30 days (57% vs 50%) as well as after 90 days (51% vs 40%). Conclusion: An unexpected, but significant, decrease in 30-day and 90-day survival was observed during the second surge in our cohort of elderly ICU patients. The reason for this is unclear. Our main concern is whether the widespread changes in practice and treatment of COVID-19 between the two surges have contributed to this increased mortality in elderly patients. Further studies are urgently warranted to provide more evidence for current practice in elderly patients. Trial registration number: NCT04321265 , registered March 19th, 2020.
Article
Full-text available
Purpose: Lactate is an established prognosticator in critical care. However, there still is insufficient evidence about its role in predicting outcome in COVID-19. This is of particular concern in older patients who have been mostly affected during the initial surge in 2020. Methods: This prospective international observation study (The COVIP study) recruited patients aged 70 years or older (ClinicalTrials.gov ID: NCT04321265) admitted to an intensive care unit (ICU) with COVID-19 disease from March 2020 to February 2021. In addition to serial lactate values (arterial blood gas analysis), we recorded several parameters, including SOFA score, ICU procedures, limitation of care, ICU- and 3-month mortality. A lactate concentration ≥ 2.0 mmol/L on the day of ICU admission (baseline) was defined as abnormal. The primary outcome was ICU-mortality. The secondary outcomes 30-day and 3-month mortality. Results: In total, data from 2860 patients were analyzed. In most patients (68%), serum lactate was lower than 2 mmol/L. Elevated baseline serum lactate was associated with significantly higher ICU- and 3-month mortality (53% vs. 43%, and 71% vs. 57%, respectively, p < 0.001). In the multivariable analysis, the maximum lactate concentration on day 1 was independently associated with ICU mortality (aOR 1.06 95% CI 1.02-1.11; p = 0.007), 30-day mortality (aOR 1.07 95% CI 1.02-1.13; p = 0.005) and 3-month mortality (aOR 1.15 95% CI 1.08-1.24; p < 0.001) after adjustment for age, gender, SOFA score, and frailty. In 826 patients with baseline lactate ≥ 2 mmol/L sufficient data to calculate the difference between maximal levels on days 1 and 2 (∆ serum lactate) were available. A decreasing lactate concentration over time was inversely associated with ICU mortality after multivariate adjustment for SOFA score, age, Clinical Frailty Scale, and gender (aOR 0.60 95% CI 0.42-0.85; p = 0.004). Conclusion: In critically ill old intensive care patients suffering from COVID-19, lactate and its kinetics are valuable tools for outcome prediction. Trial registration number: NCT04321265.
Article
Full-text available
Objective To describe patient characteristics, symptoms, patterns of care and outcomes for patients hospitalised with COVID-19 in Michigan. Design Multicentre retrospective cohort study. Setting 32 acute care hospitals in the state of Michigan. Participants Patients discharged (16 March–11 May 2020) with suspected or confirmed COVID-19 were identified. Trained abstractors collected demographic information on all patients and detailed clinical data on a subset of COVID-19-positive patients. Primary outcome measurements Patient characteristics, treatment and outcomes including cardiopulmonary resuscitation, mortality and venous thromboembolism within and across hospitals. Results Demographic-only data from 1593 COVID-19-positive and 1259 persons under investigation discharges were collected. Among 1024 cases with detailed data, the median age was 63 years; median body mass index was 30.6; and 51.4% were black. Cough, fever and shortness of breath were the top symptoms. 37.2% reported a known COVID-19 contact; 7.0% were healthcare workers; and 16.1% presented from congregated living facilities. During hospitalisation, 232 (22.7%) patients were treated in an intensive care unit (ICU); 558 (54.9%) in a ‘cohorted’ unit; 161 (15.7%) received mechanical ventilation; and 90 (8.8%) received high-flow nasal cannula. ICU patients more often received hydroxychloroquine (66% vs 46%), corticosteroids (34% vs 18%) and antibiotic therapy (92% vs 71%) than general ward patients (p<0.05 for all). Overall, 219 (21.4%) patients died, with in-hospital mortality ranging from 7.9% to 45.7% across hospitals. 73% received at least one COVID-19-specific treatment, ranging from 32% to 96% across sites. Across 14 hospitals, the proportion of patients admitted directly to an ICU ranged from 0% to 43.8%; mechanical ventilation on admission from 0% to 12.8%; mortality from 7.9% to 45.7%. Use of at least one COVID-19-specific therapy varied from 32% to 96.3% across sites. Conclusions During the early days of the Michigan outbreak of COVID-19, patient characteristics, treatment and outcomes varied widely within and across hospitals.
Article
Full-text available
As predicting the trajectory of COVID-19 is challenging, machine learning models could assist physicians in identifying high-risk individuals. This study compares the performance of 18 machine learning algorithms for predicting ICU admission and mortality among COVID-19 patients. Using COVID-19 patient data from the Mass General Brigham (MGB) Healthcare database, we developed and internally validated models using patients presenting to the Emergency Department (ED) between March-April 2020 (n = 3597) and further validated them using temporally distinct individuals who presented to the ED between May-August 2020 (n = 1711). We show that ensemble-based models perform better than other model types at predicting both 5-day ICU admission and 28-day mortality from COVID-19. CRP, LDH, and O2 saturation were important for ICU admission models whereas eGFR <60 ml/min/1.73m2, and neutrophil and lymphocyte percentages were the most important variables for predicting mortality. Implementing such models could help in clinical decision-making for future infectious disease outbreaks including COVID-19.
Article
Full-text available
Background: An accurate prediction of COVID-19 patient disease severity would greatly improve care delivery and resource allocation, and thereby reduce mortality risks, especially in less developed countries. There are many patient-related factors, such as pre-existing comorbidities that affect disease severity that could be used to aid prediction. Objective: Since rapid automated profiling of peripheral blood samples is widely available, we investigated how such data from the peripheral blood of COVID-19 patients might be used to predict clinical outcomes. Methods: We thus investigated such clinical datasets from COVID-19 patients with known outcomes by combining statistical comparison and correlation methods with machine learning algorithms; the latter included decision tree, random forest, variants of gradient boosting machine, support vector machine, K-nearest neighbour and deep learning methods. Results: Our work revealed several clinical parameters measurable in blood samples as factors that can discriminate between healthy people and COVID-19 positive patients, and showed their value in predicting later severity of COVID-19 symptoms. We thus developed a number of analytical methods that showed accuracy and precision scores for disease severity predictions as above 90%. Conclusions: We developed methodologies to analyse patient routine clinical data which enables more accurate prediction of COVID-19 patient outcomes. This type of approach could, by employing standard hospital laboratory analyses of patient blood, be utilised to identify COVID-19 patients at high risk of mortality and so enable optimised hospital facility for COVID-19 treatment. Clinicaltrial:
Article
Full-text available
Background The COVID-19 pandemic has led highly developed healthcare systems to the brink of collapse due to the large numbers of patients being admitted into hospitals. One of the potential prognostic indicators in patients with COVID-19 is frailty. The degree of frailty could be used to assist both the triage into intensive care, and decisions regarding treatment limitations. Our study sought to determine the interaction of frailty and age in elderly COVID-19 ICU patients. Methods A prospective multicentre study of COVID-19 patients ≥ 70 years admitted to intensive care in 138 ICUs from 28 countries was conducted. The primary endpoint was 30-day mortality. Frailty was assessed using the clinical frailty scale. Additionally, comorbidities, management strategies and treatment limitations were recorded. Results The study included 1346 patients (28% female) with a median age of 75 years (IQR 72–78, range 70–96), 16.3% were older than 80 years, and 21% of the patients were frail. The overall survival at 30 days was 59% (95% CI 56–62), with 66% (63–69) in fit, 53% (47–61) in vulnerable and 41% (35–47) in frail patients ( p < 0.001). In frail patients, there was no difference in 30-day survival between different age categories. Frailty was linked to an increased use of treatment limitations and less use of mechanical ventilation. In a model controlling for age, disease severity, sex, treatment limitations and comorbidities, frailty was independently associated with lower survival. Conclusion Frailty provides relevant prognostic information in elderly COVID-19 patients in addition to age and comorbidities. Trial registration Clinicaltrials.gov: NCT04321265 , registered 19 March 2020.
Article
Full-text available
Background: The pandemic caused by the SARS-Cov2 virus will probably stand as the greatest health catastrophe of the modern era. The Spanish healthcare system has been exposed to uncontrollable numbers of patients in a short period of time, causing system collapse. Given that diagnosis is not immediate and there is no effective treatment, other tools have had to be developed to identify patients at risk of severe disease complications, and thus optimize material and human resources in health care. There are no tools to establish which patients have a worse prognosis than others. Objective: In this study, we aimed to process a sample of electronic health records of COVID-19 patients in order to develop a machine learning model to predict the severity of infection and mortality through clinical laboratory parameters. Early patient classification can help optimize material and human resources, and analysis of the most important features of the model could provide insights into the disease. Methods: After an initial performance evaluation based on a comparison with several other well-known methods, the extreme gradient boosting (XGBoost) algorithm was chosen as the predictive method for this study. In addition, SHAP (SHapley Additive exPlanations) was used to analyze the importance of the features of the resulting model. Results: After data preprocessing, 1823 confirmed COVID-19 patients and 32 predictor features were selected. On bootstrap validation, the XGBoost classifier yielded a value of 0.97 (95% CI 0.96-0.98) for the area under the receiver operator characteristic curve, 0.86 (95% CI 0.80-0.91) for the area under the precision-recall curve, 0.94 (95% CI 0.92-0.95) for accuracy, 0.77 (95% CI 0.72-0.83) for F-score, 0.93 (95% CI 0.89-0.98) for sensitivity, and 0.91 (95% CI 0.86-0.96) for specificity. The four most relevant features for model prediction were LDH, C-reactive protein, neutrophils, and urea. Conclusions: The predictive model obtained in this work achieved excellent results in the discrimination of COVID-19 dead patients, by mainly employing laboratory parameter values. The analysis of the resulting model identified a set of features with the most significant impact on the prediction, and so relating them to a higher risk of mortality.
Article
Full-text available
Background: Predicting early respiratory failure in COVID-19 can help triage patients to higher levels of care, allocate scarce resources, and reduce morbidity and mortality by appropriately monitoring and treating patients at greatest risk for deterioration. Given the complexity of COVID-19 disease, machine learning (ML) approaches may support clinical decision making for patients with this disease. Objective: Our objective is to derive a machine learning model that predicts respiratory failure within 48 hours of admission based on data from the emergency department (ED). Methods: Data was collected from patients with COVID-19 who were admitted to Northwell Health acute care hospitals and discharged, died, or spent a minimum of 48 hours in the hospital between March 1, 2020 and May 11, 2020. Of 11,525 patients, 933 (8.1%) were placed on invasive mechanical ventilation within 48 hours of admission. Variables used by the models included clinical and laboratory data commonly collected in the ED. We trained and validated three predictive models (two based on XGBoost, one that utilized logistic regression) using cross hospitals validation. We compared model performance between all three models as well as an established early warning score (Modified Early Warning Score (MEWS)) using receiver operating characteristic (ROC) curves, precision-recall (PR) curves, and other metrics. Results: The XGBoost model had the highest mean accuracy of 0.919 (AUC = 0.77), outperforming the other two models as well as MEWS. Important predictor variables included the type of oxygen delivery used in the ED, patient age, Emergency Severity Index (ESI), respiratory rate, serum lactate, and demographic characteristics. Conclusions: XGBoost has high predictive accuracy, outperforming other early warning scores. The clinical plausibility and predictive ability of XGBoost suggest that the model could be used to predict 48-hour respiratory failure in admitted patients with COVID-19. Clinicaltrial:
Article
Full-text available
The role of physicians has always been to synthesize the data available to them to identify diagnostic patterns that guide treatment and follow response. Today, increasingly sophisticated machine learning algorithms may grow to support clinical experts in some of these tasks. Machine learning has the potential to benefit patients and cardiologists, but only if clinicians take an active role in bringing these new algorithms into practice. The aim of this review is to introduce clinicians who are not data science experts to key concepts in machine learning that will allow them to better understand the field and evaluate new literature and developments. The current published data in machine learning for cardiovascular disease is then summarized, using both a bibliometric survey, with code publicly available to enable similar analysis for any research topic of interest, and select case studies. Finally, several ways that clinicians can and must be involved in this emerging field are presented.
Article
Full-text available
Background : The unpredictability of the progression of coronavirus disease 2019 (COVID-19) may be attributed to the low precision of the tools used to predict the prognosis of this disease. Objective : To identify the predictors associated with poor clinical outcomes in patients with COVID-19. Methods : Relevant articles from PubMed, Embase, Cochrane, and Web of Science were searched as of April 5, 2020. The quality of the included papers was appraised using the Newcastle-Ottawa scale (NOS). Data of interest were collected and evaluated for their compatibility for the meta-analysis. Cumulative calculations to determine the correlation and effect estimates were performed using the Z test. Results : In total, 19 papers recording 1,934 mild and 1,644 severe cases of COVID-19 were included. Based on the initial evaluation, 62 potential risk factors were identified for the meta-analysis. Several comorbidities, including chronic respiratory disease, cardiovascular disease, diabetes mellitus, and hypertension were observed more frequent among patients with severe COVID-19 than with the mild ones. Compared to the mild form, severe COVID-19 was associated with symptoms such as dyspnea, anorexia, fatigue, increased respiratory rate, and high systolic blood pressure. Lower levels of lymphocytes and hemoglobin; elevated levels of leukocytes, aspartate aminotransferase, alanine aminotransferase, blood creatinine, blood urea nitrogen, high-sensitivity troponin, creatine kinase, high-sensitivity C-reactive protein, interleukin 6, D-dimer, ferritin, lactate dehydrogenase, and procalcitonin; and a high erythrocyte sedimentation rate were also associated with severe COVID-19. Conclusion : More than 30 risk factors are associated with a higher risk of severe COVID-19. These may serve as useful baseline parameters in the development of prediction tools for COVID-19 prognosis.
Article
Full-text available
Background: Machine learning (ML) models require large datasets which may be siloed across different healthcare institutions. Current ML studies focusing on coronavirus disease 2019 (COVID-19) are limited to single hospital data which limits model generalizability. Objective: Using federated learning, a ML technique that avoids locally aggregating raw clinical data across multiple institutions, we predict mortality within seven days in hospitalized COVID-19 patients. Methods: Patient data was collected from Electronic Health Records (EHRs) from five hospitals within the Mount Sinai Health System (MSHS). Logistic Regression with L1 regularization (LASSO) and Multilayer Perceptron (MLP) models were trained using local data at each site, a pooled model with combined data from all five sites, and a federated model that only shared parameters with a central aggregator. Results: LASSO-federated outperformed LASSO-local at three hospitals, and MLP-federated performed better than MLP-local at all five hospitals as measured by area under the receiver-operating characteristic (AUC-ROC). LASSO-pooled outperformed LASSO-federated at all hospitals, and MLP-federated outperformed MLP-pooled at two hospitals. Conclusions: Federated learning shows promise in COVID-19 EHR data to develop robust predictive models without compromising patient privacy. Clinicaltrial:
Article
Full-text available
Background Patients with COVID-19 in the intensive care unit (ICU) have a high mortality rate, and methods to assess patients’ prognosis early and administer precise treatment are of great significance. Objective The aim of this study was to use machine learning to construct a model for the analysis of risk factors and prediction of mortality among ICU patients with COVID-19. Methods In this study, 123 patients with COVID-19 in the ICU of Vulcan Hill Hospital were retrospectively selected from the database, and the data were randomly divided into a training data set (n=98) and test data set (n=25) with a 4:1 ratio. Significance tests, correlation analysis, and factor analysis were used to screen 100 potential risk factors individually. Conventional logistic regression methods and four machine learning algorithms were used to construct the risk prediction model for the prognosis of patients with COVID-19 in the ICU. The performance of these machine learning models was measured by the area under the receiver operating characteristic curve (AUC). Interpretation and evaluation of the risk prediction model were performed using calibration curves, SHapley Additive exPlanations (SHAP), Local Interpretable Model-Agnostic Explanations (LIME), etc, to ensure its stability and reliability. The outcome was based on the ICU deaths recorded from the database. Results Layer-by-layer screening of 100 potential risk factors finally revealed 8 important risk factors that were included in the risk prediction model: lymphocyte percentage, prothrombin time, lactate dehydrogenase, total bilirubin, eosinophil percentage, creatinine, neutrophil percentage, and albumin level. Finally, an eXtreme Gradient Boosting (XGBoost) model established with the 8 important risk factors showed the best recognition ability in the training set of 5-fold cross validation (AUC=0.86) and the verification queue (AUC=0.92). The calibration curve showed that the risk predicted by the model was in good agreement with the actual risk. In addition, using the SHAP and LIME algorithms, feature interpretation and sample prediction interpretation algorithms of the XGBoost black box model were implemented. Additionally, the model was translated into a web-based risk calculator that is freely available for public usage. Conclusions The 8-factor XGBoost model predicts risk of death in ICU patients with COVID-19 well; it initially demonstrates stability and can be used effectively to predict COVID-19 prognosis in ICU patients.
Article
Full-text available
Background: The novel coronavirus responsible for COVID-19 has caused havoc with patients presenting a spectrum of complications forcing the healthcare experts around the globe to explore new technological solutions, and treatment plans. Artificial Intelligence (AI) based technologies have played a substantial role in solving complex problems, and several organizations have been swift to adopt and customize them in response to the challenges posed by the COVID-19 pandemic. Objective: The objective of this study is to conduct a systematic literature review on the role of AI as a comprehensive and decisive technology to fight the COVID-19 crisis in the arena of epidemiology, diagnosis, and disease progression. Methods: A systematic search in PubMed, Web of Science, and CINAHL databases was performed according to the Preferred Reporting Items for Systematic Reviews and Meta-analysis (PRISMA) guidelines to identify all potentially relevant studies published and made available between December 1, 2019, and June 27, 2020. The search syntax was built using keywords specific to COVID-19 and AI. Results: The search strategy resulted in 419 articles, published and made available between December 1, 2019, and June 27, 2020. Of which, 130 publications were selected for analysis. The studies were classified into three themes based on AI applications employed to combat the COVID-19 crisis: Computational Epidemiology (CE), Early Detection and Diagnosis (EDD), and Disease Progression (DP). Of the 130 studies, 71 focused on predicting the outbreak, the impact of containment policies, and potential drug discoveries, which were grouped into the CE theme. For the EDD, we grouped forty studies that applied AI techniques to detect the presence of COVID-19 using the patients' radiological images or lab results. Nineteen publications that focused on predicting the disease progression, outcomes (recovery and mortality), Length of Stay (LOS), and number of Intensive Care Unit (ICU) days for COVID-19 positive patients were classified under the DP theme. Conclusions: In this systematic review, we assembled the current COVID-19 literature that utilized AI methods to provide insights into the COVID-19 themes, highlighting the important variables, data types, and available COVID-19 resources that can assist in facilitating clinical and translational research. Clinicaltrial:
Article
Full-text available
Objective: To assess the consistency of machine learning and statistical techniques in predicting individual level and population level risks of cardiovascular disease and the effects of censoring on risk predictions. Design: Longitudinal cohort study from 1 January 1998 to 31 December 2018. Setting and participants: 3.6 million patients from the Clinical Practice Research Datalink registered at 391 general practices in England with linked hospital admission and mortality records. Main outcome measures: Model performance including discrimination, calibration, and consistency of individual risk prediction for the same patients among models with comparable model performance. 19 different prediction techniques were applied, including 12 families of machine learning models (grid searched for best models), three Cox proportional hazards models (local fitted, QRISK3, and Framingham), three parametric survival models, and one logistic model. Results: The various models had similar population level performance (C statistics of about 0.87 and similar calibration). However, the predictions for individual risks of cardiovascular disease varied widely between and within different types of machine learning and statistical models, especially in patients with higher risks. A patient with a risk of 9.5-10.5% predicted by QRISK3 had a risk of 2.9-9.2% in a random forest and 2.4-7.2% in a neural network. The differences in predicted risks between QRISK3 and a neural network ranged between -23.2% and 0.1% (95% range). Models that ignored censoring (that is, assumed censored patients to be event free) substantially underestimated risk of cardiovascular disease. Of the 223 815 patients with a cardiovascular disease risk above 7.5% with QRISK3, 57.8% would be reclassified below 7.5% when using another model. Conclusions: A variety of models predicted risks for the same patients very differently despite similar model performances. The logistic models and commonly used machine learning models should not be directly applied to the prediction of long term risks without considering censoring. Survival models that consider censoring and that are explainable, such as QRISK3, are preferable. The level of consistency within and between models should be routinely assessed before they are used for clinical decision making.
Article
Full-text available
Background Prioritizing patients in need of intensive care is necessary to reduce the mortality rate during the COVID-19 pandemic. Although several scoring methods have been introduced, many require laboratory or radiographic findings that are not always easily available. Objective The purpose of this study was to develop a machine learning model that predicts the need for intensive care for patients with COVID-19 using easily obtainable characteristics—baseline demographics, comorbidities, and symptoms. Methods A retrospective study was performed using a nationwide cohort in South Korea. Patients admitted to 100 hospitals from January 25, 2020, to June 3, 2020, were included. Patient information was collected retrospectively by the attending physicians in each hospital and uploaded to an online case report form. Variables that could be easily provided were extracted. The variables were age, sex, smoking history, body temperature, comorbidities, activities of daily living, and symptoms. The primary outcome was the need for intensive care, defined as admission to the intensive care unit, use of extracorporeal life support, mechanical ventilation, vasopressors, or death within 30 days of hospitalization. Patients admitted until March 20, 2020, were included in the derivation group to develop prediction models using an automated machine learning technique. The models were externally validated in patients admitted after March 21, 2020. The machine learning model with the best discrimination performance was selected and compared against the CURB-65 (confusion, urea, respiratory rate, blood pressure, and 65 years of age or older) score using the area under the receiver operating characteristic curve (AUC). Results A total of 4787 patients were included in the analysis, of which 3294 were assigned to the derivation group and 1493 to the validation group. Among the 4787 patients, 460 (9.6%) patients needed intensive care. Of the 55 machine learning models developed, the XGBoost model revealed the highest discrimination performance. The AUC of the XGBoost model was 0.897 (95% CI 0.877-0.917) for the derivation group and 0.885 (95% CI 0.855-0.915) for the validation group. Both the AUCs were superior to those of CURB-65, which were 0.836 (95% CI 0.825-0.847) and 0.843 (95% CI 0.829-0.857), respectively. Conclusions We developed a machine learning model comprising simple patient-provided characteristics, which can efficiently predict the need for intensive care among patients with COVID-19.
Article
Full-text available
Background Many factors involved in the onset and clinical course of the ongoing COVID-19 pandemic are still unknown. Although big data analytics and artificial intelligence are widely used in the realms of health and medicine, researchers are only beginning to use these tools to explore the clinical characteristics and predictive factors of patients with COVID-19. Objective Our primary objectives are to describe the clinical characteristics and determine the factors that predict intensive care unit (ICU) admission of patients with COVID-19. Determining these factors using a well-defined population can increase our understanding of the real-world epidemiology of the disease. Methods We used a combination of classic epidemiological methods, natural language processing (NLP), and machine learning (for predictive modeling) to analyze the electronic health records (EHRs) of patients with COVID-19. We explored the unstructured free text in the EHRs within the Servicio de Salud de Castilla-La Mancha (SESCAM) Health Care Network (Castilla-La Mancha, Spain) from the entire population with available EHRs (1,364,924 patients) from January 1 to March 29, 2020. We extracted related clinical information regarding diagnosis, progression, and outcome for all COVID-19 cases. Results A total of 10,504 patients with a clinical or polymerase chain reaction–confirmed diagnosis of COVID-19 were identified; 5519 (52.5%) were male, with a mean age of 58.2 years (SD 19.7). Upon admission, the most common symptoms were cough, fever, and dyspnea; however, all three symptoms occurred in fewer than half of the cases. Overall, 6.1% (83/1353) of hospitalized patients required ICU admission. Using a machine-learning, data-driven algorithm, we identified that a combination of age, fever, and tachypnea was the most parsimonious predictor of ICU admission; patients younger than 56 years, without tachypnea, and temperature <39 degrees Celsius (or >39 ºC without respiratory crackles) were not admitted to the ICU. In contrast, patients with COVID-19 aged 40 to 79 years were likely to be admitted to the ICU if they had tachypnea and delayed their visit to the emergency department after being seen in primary care. Conclusions Our results show that a combination of easily obtainable clinical variables (age, fever, and tachypnea with or without respiratory crackles) predicts whether patients with COVID-19 will require ICU admission.
Article
Full-text available
Background: The unpredictability of the progression of coronavirus disease 2019 (COVID-19) may be attributed to the low precision of the tools used to predict the prognosis of this disease. Objective: To identify the predictors associated with poor clinical outcomes in patients with COVID-19. Methods: Relevant articles from PubMed, Embase, Cochrane, and Web of Science were searched and extracted as of April 5, 2020. Data of interest were collected and evaluated for their compatibility for the meta-analysis. Cumulative calculations to determine the correlation and effect estimates were performed using the Z test. Results: In total, 19 papers recording 1,934 mild and 1,644 severe cases of COVID-19 were included. Based on the initial evaluation, 62 potential risk factors were identified for the meta-analysis. Several comorbidities, including chronic respiratory disease, cardiovascular disease, diabetes mellitus, and hypertension were observed more frequent among patients with severe COVID-19 than with the mild ones. Compared to the mild form, severe COVID-19 was associated with symptoms such as dyspnea, anorexia, fatigue, increased respiratory rate, and high systolic blood pressure. Lower levels of lymphocytes and hemoglobin; elevated levels of leukocytes, aspartate aminotransferase, alanine aminotransferase, blood creatinine, blood urea nitrogen, high-sensitivity troponin, creatine kinase, high-sensitivity C-reactive protein, interleukin 6, D-dimer, ferritin, lactate dehydrogenase, and procalcitonin; and a high erythrocyte sedimentation rate were also associated with severe COVID-19. Conclusion: More than 30 risk factors are associated with a higher risk of severe COVID-19. These may serve as useful baseline parameters in the development of prediction tools for COVID-19 prognosis.
Article
Full-text available
This study aimed to develop risk scores based on clinical characteristics at presentation to predict intensive care unit (ICU) admission and mortality in COVID-19 patients. 641 hospitalized patients with laboratory-confirmed COVID-19 were selected from 4997 persons under investigation. We performed a retrospective review of medical records of demographics, comorbidities and laboratory tests at the initial presentation. Primary outcomes were ICU admission and death. Logistic regression was used to identify independent clinical variables predicting the two outcomes. The model was validated by splitting the data into 70% for training and 30% for testing. Performance accuracy was evaluated using area under the curve (AUC) of the receiver operating characteristic analysis (ROC). Five significant variables predicting ICU admission were lactate dehydrogenase, procalcitonin, pulse oxygen saturation, smoking history, and lymphocyte count. Seven significant variables predicting mortality were heart failure, procalcitonin, lactate dehydrogenase, chronic obstructive pulmonary disease, pulse oxygen saturation, heart rate, and age. The mortality group uniquely contained cardiopulmonary variables. The risk score model yielded good accuracy with an AUC of 0.74 ([95% CI, 0.63–0.85], p = 0.001) for predicting ICU admission and 0.83 ([95% CI, 0.73–0.92], p<0.001) for predicting mortality for the testing dataset. This study identified key independent clinical variables that predicted ICU admission and mortality associated with COVID-19. This risk score system may prove useful for frontline physicians in clinical decision-making under time-sensitive and resource-constrained environment.
Article
Full-text available
Older people have paid a huge toll in terms of mortality during the coronavirus disease-19 (COVID-19) pandemic. Frailty may have contributed to the vulnerability of older people to more severe clinical presentation. We aimed at reviewing available evidence about frailty and COVID-19. We searched PUBMED, Web of Science, and EMBASE from 1 December 2019 to 29 May 2020. Study selection and data extraction were performed by three independent reviewers. Qualitative synthesis was conducted and quantitative data extracted when available. Forty papers were included: 13 editorials, 15 recommendations/guidelines, 3 reviews, 1 clinical trial, 6 observational studies, 2 case reports. Editorials and reviews underlined the potential clinical relevance of assessing frailty among older patients with COVID-19. However, frailty was only investigated in regards to its association with overall mortality, hospital contagion, intensive care unit admission rates, and disease phenotypes in the few observational studies retrieved. Specific interventions in relation to frailty or its impact on COVID-19 treatments have not been evaluated yet. Even with such limited evidence, clinical recommendations on the use of frailty tools have been proposed to support decision making about escalation plan. Ongoing initiatives are expected to improve knowledge of COVID-19 interaction with frailty and to promote patient-centered approaches.
Article
Full-text available
The elderly may represent a specific cluster of high-risk patients for developing COVID-19 with rapidly progressive clinical deterioration. Indeed, in older individuals, immunosenescence and comorbid disorders are more likely to promote viral-induced cytokine storm resulting in life-threatening respiratory failure and multisystemic involvement. Early diagnosis and individualized therapeutic management should be developed for elderly subjects based on personal medical history and polypharmacotherapy. Our review examines the pathogenesis and clinical implications of ageing in COVID-19 patients; finally, we discuss the evidence and controversies in the management in the long-stay residential care homes and aspects of end-of-life care for elderly patients with COVID-19.
Article
Full-text available
The evolving dynamics of coronavirus disease 2019 (COVID-19) and the increasing infection numbers require diagnostic tools to identify patients at high risk for a severe disease course. Here we evaluate clinical and imaging parameters for estimating the need of intensive care unit (ICU) treatment. We collected clinical, laboratory and imaging data from 65 patients with confirmed COVID-19 infection based on polymerase chain reaction (PCR) testing. Two radiologists evaluated the severity of findings in computed tomography (CT) images on a scale from 1 (no characteristic signs of COVID-19) to 5 (confluent ground glass opacities in over 50% of the lung parenchyma). The volume of affected lung was quantified using commercially available software. Machine learning modelling was performed to estimate the risk for ICU treatment. Patients with a severe course of COVID-19 had significantly increased interleukin (IL)-6, C-reactive protein (CRP), and leukocyte counts and significantly decreased lymphocyte counts. The radiological severity grading was significantly increased in ICU patients. Multivariate random forest modelling showed a mean ± standard deviation sensitivity, specificity and accuracy of 0.72 ± 0.1, 0.86 ± 0.16 and 0.80 ± 0.1 and a receiver operating characteristic-area under curve (ROC-AUC) of 0.79 ± 0.1. The need for ICU treatment is independently associated with affected lung volume, radiological severity score, CRP, and IL-6.
Article
Full-text available
Objective To review and critically appraise published and preprint reports of prediction models for diagnosing coronavirus disease 2019 (covid-19) in patients with suspected infection, for prognosis of patients with covid-19, and for detecting people in the general population at risk of being admitted to hospital for covid-19 pneumonia. Design Rapid systematic review and critical appraisal. Data sources PubMed and Embase through Ovid, Arxiv, medRxiv, and bioRxiv up to 24 March 2020. Study selection Studies that developed or validated a multivariable covid-19 related prediction model. Data extraction At least two authors independently extracted data using the CHARMS (critical appraisal and data extraction for systematic reviews of prediction modelling studies) checklist; risk of bias was assessed using PROBAST (prediction model risk of bias assessment tool). Results 2696 titles were screened, and 27 studies describing 31 prediction models were included. Three models were identified for predicting hospital admission from pneumonia and other events (as proxy outcomes for covid-19 pneumonia) in the general population; 18 diagnostic models for detecting covid-19 infection (13 were machine learning based on computed tomography scans); and 10 prognostic models for predicting mortality risk, progression to severe disease, or length of hospital stay. Only one study used patient data from outside of China. The most reported predictors of presence of covid-19 in patients with suspected disease included age, body temperature, and signs and symptoms. The most reported predictors of severe prognosis in patients with covid-19 included age, sex, features derived from computed tomography scans, C reactive protein, lactic dehydrogenase, and lymphocyte count. C index estimates ranged from 0.73 to 0.81 in prediction models for the general population (reported for all three models), from 0.81 to more than 0.99 in diagnostic models (reported for 13 of the 18 models), and from 0.85 to 0.98 in prognostic models (reported for six of the 10 models). All studies were rated at high risk of bias, mostly because of non-representative selection of control patients, exclusion of patients who had not experienced the event of interest by the end of the study, and high risk of model overfitting. Reporting quality varied substantially between studies. Most reports did not include a description of the study population or intended use of the models, and calibration of predictions was rarely assessed. Conclusion Prediction models for covid-19 are quickly entering the academic literature to support medical decision making at a time when they are urgently needed. This review indicates that proposed models are poorly reported, at high risk of bias, and their reported performance is probably optimistic. Immediate sharing of well documented individual participant data from covid-19 studies is needed for collaborative efforts to develop more rigorous prediction models and validate existing ones. The predictors identified in included studies could be considered as candidate predictors for new models. Methodological guidance should be followed because unreliable predictions could cause more harm than benefit in guiding clinical decisions. Finally, studies should adhere to the TRIPOD (transparent reporting of a multivariable prediction model for individual prognosis or diagnosis) reporting guideline. Systematic review registration Protocol https://osf.io/ehc47/ , registration https://osf.io/wy245 .
Article
Full-text available
Background: Prognosticating the course of diseases to inform decision-making is a key component of intensive care medicine. For several applications in medicine, new methods from the field of artificial intelligence (AI) and machine learning have already outperformed conventional prediction models. Due to their technical characteristics, these methods will present new ethical challenges to the intensivist. Results: In addition to the standards of data stewardship in medicine, the selection of datasets and algorithms to create AI prognostication models must involve extensive scrutiny to avoid biases and, consequently, injustice against individuals or groups of patients. Assessment of these models for compliance with the ethical principles of beneficence and non-maleficence should also include quantification of predictive uncertainty. Respect for patients' autonomy during decision-making requires transparency of the data processing by AI models to explain the predictions derived from these models. Moreover, a system of continuous oversight can help to maintain public trust in this technology. Based on these considerations as well as recent guidelines, we propose a pathway to an ethical implementation of AI-based prognostication. It includes a checklist for new AI models that deals with medical and technical topics as well as patient- and system-centered issues. Conclusion: AI models for prognostication will become valuable tools in intensive care. However, they require technical refinement and a careful implementation according to the standards of medical ethics.
Article
Full-text available
Purpose: Very old critical ill patients are a rapid expanding group in the ICU. Indications for admission, triage criteria and level of care are frequently discussed for such patients. However, most relevant outcome studies in this group frequently find an increased mortality and a reduced quality of life in survivors. The main objective was to study the impact of frailty compared with other variables with regards to short-term outcome in the very old ICU population. Methods: A transnational prospective cohort study from October 2016 to May 2017 with 30 days follow-up was set up by the European Society of Intensive Care Medicine. In total 311 ICUs from 21 European countries participated. The ICUs included the first consecutive 20 very old (≥ 80 years) patients admitted to the ICU within a 3-month inclusion period. Frailty, SOFA score and therapeutic procedures were registered, in addition to limitations of care. For measurement of frailty the Clinical Frailty Scale was used at ICU admission. The main outcomes were ICU and 30-day mortality and survival at 30 days. Results: A total of 5021 patients with a median age of 84 years (IQR 81-86 years) were included in the final analysis, 2404 (47.9%) were women. Admission was classified as acute in 4215 (83.9%) of the patients. Overall ICU and 30-day mortality rates were 22.1% and 32.6%. During ICU stay 23.8% of the patients did not receive specific ICU procedures: ventilation, vasoactive drugs or renal replacement therapy. Frailty (values ≥ 5) was found in 43.1% and was independently related to 30-day survival (HR 1.54; 95% CI 1.38-1.73) for frail versus non-frail. Conclusions: Among very old patients (≥ 80 years) admitted to the ICU, the consecutive classes in Clinical Frailty Scale were inversely associated with short-term survival. The scale had a very low number of missing data. These findings provide support to add frailty to the clinical assessment in this patient group. Trial registration: ClinicalTrials.gov (ID: NCT03134807).
Article
Full-text available
Importance Time-limited trials of intensive care are commonly used in patients perceived to have a poor prognosis. The optimal duration of such trials is unknown. Factors such as a cancer diagnosis are associated with clinician pessimism and may affect the decision to limit care independent of a patient’s severity of illness.Objective To identify the optimal duration of intensive care for short-term mortality in critically ill patients with cancer.Design, Setting, and Participants Decision analysis using a state-transition microsimulation model was performed to simulate the hospital course of patients with poor-prognosis primary tumors, metastatic disease, or hematologic malignant neoplasms admitted to medical and surgical intensive care units. Transition probabilities were derived from 920 participants stratified by sequential organ failure assessment (SOFA) scores to identify severity of illness. The model was validated in 3 independent cohorts with 349, 158, and 117 participants from quaternary care academic hospitals. Monte Carlo microsimulation was performed, followed by probabilistic sensitivity analysis. Outcomes were assessed in the overall cohort and in solid tumors alone.Interventions Time-unlimited vs time-limited trials of intensive care.Main Outcomes and Measures 30-day all-cause mortality and mean survival duration.Results The SOFA scores at ICU admission were significantly associated with mortality. A 3-, 8-, or 15-day trial of intensive care resulted in decreased mean 30-day survival vs aggressive care in all but the sickest patients (SOFA score, 5-9: 48.4% [95% CI, 48.0%-48.8%], 60.6% [95% CI, 60.2%-61.1%], and 66.8% [95% CI, 66.4%-67.2%], respectively, vs 74.6% [95% CI, 74.3%-75.0%] with time-unlimited aggressive care; SOFA score, 10-14: 36.2% [95% CI, 35.8%-36.6%], 44.1% [95% CI, 43.6%-44.5%], and 46.1% [95% CI, 45.6%-46.5%], respectively, vs 48.4% [95% CI, 48.0%-48.8%] with aggressive care; SOFA score, ≥15: 5.8% [95% CI, 5.6%-6.0%], 8.1% [95% CI, 7.9%-8.3%], and 8.3% [95% CI, 8.1%-8.6%], respectively, vs 8.8% [95% CI, 8.5%-9.0%] with aggressive care). However, the clinical magnitude of these differences was variable. Trial durations of 8 days in the sickest patients offered mean survival duration that was no more than 1 day different from time-unlimited care, whereas trial durations of 10 to 12 days were required in healthier patients. For the subset of patients with solid tumors, trial durations of 1 to 4 days offered mean survival that was not statistically significantly different from time-unlimited care.Conclusions and Relevance Trials of ICU care lasting 1 to 4 days may be sufficient in patients with poor-prognosis solid tumors, whereas patients with hematologic malignant neoplasms or less severe illness seem to benefit from longer trials of intensive care.
Article
Full-text available
Binary classifiers are routinely evaluated with performance measures such as sensitivity and specificity, and performance is frequently illustrated with Receiver Operating Characteristics (ROC) plots. Alternative measures such as positive predictive value (PPV) and the associated Precision/Recall (PRC) plots are used less frequently. Many bioinformatics studies develop and evaluate classifiers that are to be applied to strongly imbalanced datasets in which the number of negatives outweighs the number of positives significantly. While ROC plots are visually appealing and provide an overview of a classifier's performance across a wide range of specificities, one can ask whether ROC plots could be misleading when applied in imbalanced classification scenarios. We show here that the visual interpretability of ROC plots in the context of imbalanced datasets can be deceptive with respect to conclusions about the reliability of classification performance, owing to an intuitive but wrong interpretation of specificity. PRC plots, on the other hand, can provide the viewer with an accurate prediction of future classification performance due to the fact that they evaluate the fraction of true positives among positive predictions. Our findings have potential implications for the interpretation of a large number of studies that use ROC plots on imbalanced datasets.
Article
PURPOSECritically ill elderly patients who suffer from Sars-CoV-2 disease are at high risk for organ failure. The modified MELD-XI score has not been evaluated for outcome prediction in these most vulnerable patients.METHODS The Corona Virus disease (COVID19) in Very Elderly Intensive Care Patients study (COVIP, NCT04321265) prospectively recruited patients on intensive care units (ICU), who were = 70 years. Data were collected from March 2020 to February 2021. The MELD-XI score was calculated using the highest serum bilirubin and creatinine on ICU admission. Univariate and multivariable logistic regression analyses were performed to assess associations between the MELD-XI score and mortality. The primary outcome was 30-day-mortality, the secondary outcomes were ICU- and 3-month-mortality.RESULTSIn total, data from 2,993 patients were analyzed. Most patients had a MELD-XI <12 on admission (76%). The patients with MELD-XI = 12 had a significantly higher 30-day-, ICU- and 3-month-mortality (44%vs 64%, and 42%vs. 59%, and 57%vs. 76%, p < 0.001). After adjustment for multiple confounders, MELD-XI = 12 remained significantly associated with 30-day- (aOR 1.572, CI 1.268-1.949, p < 0.001), ICU-, and 3-month-mortality.CONCLUSION In critically ill elderly intensive care patients with COVID-19, the MELD-XI score constitutes a valuable tool for an early outcome prediction.
Article
Background Objective and early identification of hospitalized patients, and particularly those with novel coronavirus disease 2019 (COVID-19), who may require mechanical ventilation (MV) may aid in delivering timely treatment. Research question Can a transparent deep learning (DL) model predict the need for MV in hospitalized patients and those with COVID-19 up to 24 hours in advance? Study design and methods We trained and externally validated a transparent DL algorithm to predict the future need for MV in hospitalized patients, including those with COVID-19, using commonly available data in electronic health records. Additionally, commonly used clinical criteria (heart rate, oxygen saturation, respiratory rate, FiO2 and pH) were used to assess future need for MV. Performance of the algorithm was evaluated using the area under receiver operating characteristic curve (AUC), sensitivity, specificity and positive predictive value. RESULTS We obtained data from over 30,000 ICU patients (including over 700 patients with COVID-19) from two academic medical centers. The performance of the model with a 24-hour prediction horizon at the development and validation sites was comparable (AUC of 0.895 versus 0.882, respectively), providing significant improvement over traditional clinical criteria (p<0.001). Prospective validation of the algorithm among patients with COVID-19 yielded AUCs in the range 0.918-0.943. Interpretation A transparent DL algorithm improves on traditional clinical criteria to predict the need for MV in hospitalized patients, including in those with COVID-19. Such an algorithm may help clinicians optimize timing of tracheal intubation, better allocate resources and staff, and improve patient care.
Article
Background: The percentage of patients in intensive care who are 80 years old or older is continually increasing. Such patients already made up more than 20% of all patients in intensive care in Germany in the years 2007-2011. Meanwhile, effective treatments that support the organs of the body and keep severely ill patients alive are also being continually developed and refined. Frailty is a key prognostic parameter. The scientifically based assessment of frailty can be highly useful in intensive care medicine with regard to consented decision-making, individualized prognostication, treatment planning, and aftercare. Methods: Pertinent publications were retrieved by a selective search in the PubMed database. On the basis of the literature assessment, a variety of screening instruments were used to assess frailty and its significance for very old, critically ill patients in German intensive care units. Results: Only a small number of screening instruments are suitable for routine use in German intensive care units. The scores vary in diagnostic precision. The Clinical Frailty Scale (CFS) enables highly accurate prognostication; it considers the patient in relation to his or her social environment, and to the reference population. Categorization is achieved by means of pictograms that are supplemented with brief written descriptions. The CFS can be used prospectively and is easy to learn. Its interrater reliability is high (weighted Cohen's κ: 0.85 [0.84; 0.87]), and it has been validated for routine use in intensive care units in Germany. Conclusion: None of the available scores enable perfect prognostication. In Germany, frailty in intensive-care patients is currently best assessed on a simple visual scale (CFS).
Article
Purpose : To evaluate the application of machine learning methods, specifically Deep Neural Networks (DNN) models for intensive care (ICU) mortality prediction. The aim was to predict mortality within 96 hours after admission to mirror the clinical situation of patient evaluation after an ICU trial, which consists of 24-48 hours of ICU treatment and then “re-triage”. The input variables were deliberately restricted to ABG values to maximise real-world practicability. Methods : We retrospectively evaluated septic patients in the multi-centre eICU dataset as well as single centre MIMIC-III dataset. Included were all patients alive after 48 hours with available data on ABG (n = 3979 and n = 9655 ICU stays for the multi-centre and single centre respectively). The primary endpoint was 96 -h-mortality. Results : The model was developed using long short-term memory (LSTM), a type of DNN designed to learn temporal dependencies between variables. Input variables were all ABG values within the first 48 hours. The SOFA score (AUC of 0.72) was moderately predictive. Logistic regression showed good performance (AUC of 0.82). The best performance was achieved by the LSTM-based model with AUC of 0.88 in the multi-centre study and AUC of 0.85 in the single centre study. Conclusions : An LSTM-based model could help physicians with the “re-triage” and the decision to restrict treatment in patients with a poor prognosis.
Article
Purpose Premorbid conditions affect prognosis of acutely-ill aged patients. Several lines of evidence suggest geriatric syndromes need to be assessed but little is known on their relative effect on the 30-day survival after ICU admission. The primary aim of this study was to describe the prevalence of frailty, cognition decline and activity of daily life in addition to the presence of comorbidity and polypharmacy and to assess their influence on 30-day survival. Methods Prospective cohort study with 242 ICUs from 22 countries. Patients 80 years or above acutely admitted over a six months period to an ICU between May 2018 and May 2019 were included. In addition to common patients’ characteristics and disease severity, we collected information on specific geriatric syndromes as potential predictive factors for 30-day survival, frailty (Clinical Frailty scale) with a CFS > 4 defining frail patients, cognitive impairment (informant questionnaire on cognitive decline in the elderly (IQCODE) with IQCODE ≥ 3.5 defining cognitive decline, and disability (measured the activity of daily life with the Katz index) with ADL ≤ 4 defining disability. A Principal Component Analysis to identify co-linearity between geriatric syndromes was performed and from this a multivariable model was built with all geriatric information or only one: CFS, IQCODE or ADL. Akaike’s information criterion across imputations was used to evaluate the goodness of fit of our models. Results We included 3920 patients with a median age of 84 years (IQR: 81–87), 53.3% males). 80% received at least one organ support. The median ICU length of stay was 3.88 days (IQR: 1.83–8). The ICU and 30-day survival were 72.5% and 61.2% respectively. The geriatric conditions were median (IQR): CFS: 4 (3–6); IQCODE: 3.19 (3–3.69); ADL: 6 (4–6); Comorbidity and Polypharmacy score (CPS): 10 (7–14). CFS, ADL and IQCODE were closely correlated. The multivariable analysis identified predictors of 1-month mortality (HR; 95% CI): Age (per 1 year increase): 1.02 (1.–1.03, p = 0.01), ICU admission diagnosis, sequential organ failure assessment score (SOFA) (per point): 1.15 (1.14–1.17, p < 0.0001) and CFS (per point): 1.1 (1.05–1.15, p < 0.001). CFS remained an independent factor after inclusion of life-sustaining treatment limitation in the model. Conclusion We confirm that frailty assessment using the CFS is able to predict short-term mortality in elderly patients admitted to ICU. Other geriatric syndromes do not add improvement to the prediction model. Since CFS is easy to measure, it should be routinely collected for all elderly ICU patients in particular in connection to advance care plans, and should be used in decision making.
Article
Purpose Changes of lactate concentration over time were reported to be associated with survival in septic patients. We aimed to evaluate delta-lactate (ΔLac) 24 h after admission (Δ24Lac) to an intensive care unit (ICU) in critically ill patients for short- and long-term prognostic relevance. Methods In total, 26,285 lactate measurements of 2191 patients admitted to a German ICU were analyzed. Inclusion criterion was a lactate concentration at admission above 2.0 mmol/L. Maximum lactate concentrations of day 1 and day 2 were used to calculate Δ24Lac. Follow-up of patients was performed retrospectively. Association of Δ24Lac and both in-hospital and long-term mortality were investigated. An optimal cut-off was calculated by means of the Youden index. Results Patients with lower Δ24Lac were of similar age, but clinically sicker. As continuous variable, higher Δ24Lac was associated with decreased in-hospital mortality (per 1% Δ24Lac; HR 0.987 95%CI 0.985–0.990; p < 0.001) and an optimal Δ24Lac cut-off was calculated at 19%. Δ24Lac ≤ 19% was associated with both increased in-hospital (15% vs 43%; OR 4.11; 95%CI 3.23–5.21; p < 0.001) and long-term mortality (HR 1.54 95%CI 1.28–1.87; p < 0.001), even after correction for APACHE II, need for catecholamines and intubation. We matched 256 patients with Δ24Lac ≤ 19% to case–controls > 19% corrected for APACHE II scores, baseline lactate level and sex: Δ24Lac ≤ 19% remained associated with lower in-hospital and long-term survival. Conclusions Lower Δ24Lac was robustly associated with adverse outcome in critically ill patients, even after correction for confounders. Δ24Lac might constitute an independent, easily available and important parameter for risk stratification in the critically ill.
Article
In critically ill patients, it is frequently challenging to identify who will benefit from admission to the intensive care unit and life-sustaining interventions when the chances of a meaningful outcome are unclear. In addition, the acute illness not only affects the patients but also family members or surrogates who often are overwhelmed and unable to make thoughtful decisions. In these circumstances, a time-limited trial (TLT) of intensive care treatment can be helpful. A TLT is an agreement to initiate all necessary treatments or treatments with clearly delineated limitations for a certain period of time to gain a more realistic understanding of the patient’s chances of a meaningful recovery or to ascertain the patient’s wishes and values. In this article, we discuss current research on different aspects of TLTs in the intensive care unit. We propose how and when to use TLTs, discuss how much time should be taken for a TLT, give an overview of the potential impact of TLTs on healthcare resources, describe ethical challenges concerning TLTs, and discuss how to evaluate a TLT.
Conference Paper
Tree boosting is a highly effective and widely used machine learning method. In this paper, we describe a scalable end-to-end tree boosting system called XGBoost, which is used widely by data scientists to achieve state-of-the-art results on many machine learning challenges. We propose a novel sparsity-aware algorithm for sparse data and weighted quantile sketch for approximate tree learning. More importantly, we provide insights on cache access patterns, data compression and sharding to build a scalable tree boosting system. By combining these insights, XGBoost scales beyond billions of examples using far fewer resources than existing systems.
Conference Paper
Decision trees are attractive classifiers due to their high execution speed. But trees derived with traditional methods often cannot be grown to arbitrary complexity for possible loss of generalization accuracy on unseen data. The limitation on complexity usually means suboptimal accuracy on training data. Following the principles of stochastic modeling, we propose a method to construct tree-based classifiers whose capacity can be arbitrarily expanded for increases in accuracy for both training and unseen data. The essence of the method is to build multiple trees in randomly selected subspaces of the feature space. Trees in, different subspaces generalize their classification in complementary ways, and their combined classification can be monotonically improved. The validity of the method is demonstrated through experiments on the recognition of handwritten digits
Of Anaesthesia and Intensive Care, Ålesund Hospital, Ålesund, Norway. Dep. of Circulation and medical imaging
  • Dep
Dep. Of Anaesthesia and Intensive Care, Ålesund Hospital, Ålesund, Norway. Dep. of Circulation and medical imaging, Norwegian university of Science and Technology, Trondheim, Norway;
This work was supported by the Forschungskommission of the Medical Faculty of the Heinrich-Heine-University Düsseldorf, No
  • J F Rrb
Corresponding author: 2018-2019, grant Agreement number 831644. This work was supported by the Forschungskommission of the Medical Faculty of the Heinrich-Heine-University Düsseldorf, No. 2018-32 to G.W. and No. 2020-21 to RRB for a Clinician Scientist Track. Author Contributions B.W., B.M., J.F., RRB, V.O. and C.J. analysed the data and wrote the first draft of the manuscript.
gave guidance and improved the paper. All authors read and approved the final manuscript. References: 1. European Society of Intensive Care, M., A. Global Sepsis, and M. Society of Critical Care, Reducing the global burden of sepsis: a positive legacy for the COVID-19 pandemic?
  • W S Mc
  • M J So
  • R M Mc
A.A. and BBP and JCS and G.W. contributed to statistical analysis and improved the paper. M.K. and M.B. and S.S. and PVH and W.S. and MC and M.E. and M.J. and SO and T.Z. and B.M. and FHA and R.M. and MC and S.L. and DL and B.G. and H.F. gave guidance and improved the paper. All authors read and approved the final manuscript. References: 1. European Society of Intensive Care, M., A. Global Sepsis, and M. Society of Critical Care, Reducing the global burden of sepsis: a positive legacy for the COVID-19 pandemic? Intensive Care Med, 2021. 47(7): p. 733-736.
  • S Lundberg
  • S.-I Lee
Lundberg, S. and S.-I. Lee, A unified approach to interpreting model predictions. arXiv preprint arXiv:1705.07874, 2017.