Electrocardiogram-based mortality prediction in patients with COVID-19 using machine learning

Abstract Funding Acknowledgements Type of funding sources: Public hospital(s). Main funding source(s): The Netherlands Organisation for Health Research and Development (ZonMw) University of Amsterdam Research Priority Area Medical Integromics OnBehalf CAPACITY-COVID19 Registry Background The electrocardiogram (ECG) is an easy to assess, widely available and inexpensive tool that is frequently used during the work-up of hospitalized COVID-19 patients. So far, no study has been conducted to evaluate if ECG-based machine learning models are able to predict all-cause in-hospital mortality in COVID-19 patients. Purpose With this study, we aim to evaluate the value of using the ECG to predict in-hospital all-cause mortality of COVID-19 patients by analyzing the ECG at hospital admission, comparing a logistic regression based approach and a DNN based approach. Secondly, we aim to identify specific ECG features associated with mortality in patients diagnosed with COVID-19. Methods and results We studied 882 patients admitted with COVID-19 across seven hospitals in the Netherlands. Raw-format 12-lead ECGs recorded after admission (<72 hours) were collected, manually assessed, and annotated using pre-defined ECG features. Using data from five out of seven centers (n = 634), two mortality prediction models were developed: (a) a logistic regression model using manually annotated ECG features, and (b) a pre-trained deep neural network (DNN) using the raw ECG waveforms. Data from two other centers (n = 248) were used for external validation. Performance of both prediction models was similar, with a mean area under the receiver operating curve of 0.69 [95%CI 0.55–0.82] for the logistic regression model and 0.71 [95%CI 0.59–0.81] for the DNN in the external validation cohort. After adjustment for age and sex, ventricular rate (OR 1.13 [95% CI 1.01–1.27] per 10 ms increase), right bundle branch block (3.26 [95% CI 1.15–9.50]), ST-depression (2.78 [95% CI 1.03–7.70]) and low QRS voltages (3.09 [95% CI 1.02-9.38]) remained as significant predictors for mortality. Conclusion This study shows that ECG-based prediction models at admission may be a valuable addition to the initial risk stratification in admitted COVID-19 patients. The DNN model showed similar performance to the logistic regression that needs time-consuming manual annotation. Several ECG features associated with mortality were identified. Figure 1: Overview of methods, using and example case: (left) logistic regression and (right) deep learning. This specific case had a high probability of in-hospital mortality (above the threshold of 30%). Follow-up of this case showed that the patient had died during admission. Abstract Figure. Overview of ML methods used

Download Full-text

Predicting the risk of inpatient hypoglycemia with machine learning using electronic health records

10.2337/figshare.12091953.v1 ◽

2020 ◽

Author(s):

Yue Ruan ◽

Alexis Bellot ◽

Zuzana Moysova ◽

Garry D. Tan ◽

Alistair Lumb ◽

...

Keyword(s):

Machine Learning ◽

Logistic Regression ◽

Regression Model ◽

Logistic Regression Model ◽

Prediction Models ◽

Machine Learning Algorithms ◽

Operating Characteristics ◽

Oxygen Saturation Level ◽

Electronic Health ◽

Clinically Significant

Objective We analyzed data from inpatients with diabetes admitted to a large university hospital to predict the risk of hypoglycemia through the use of machine learning algorithms. Research Design and Methods Four years of data was extracted from a hospital electronic health record system. This included laboratory and point-of-care blood glucose (BG) values to identify biochemical and clinically significant hypoglycaemic episodes (BG < 3.9 and < 2.9mmol/L respectively). We used patient demographics, administered medications, vital signs, laboratory results and procedures performed during the hospital stays to inform the model. Two iterations of the dataset included the doses of insulin administered and the past history of inpatient hypoglycaemia. Eighteen different prediction models were compared using the area under curve of the receiver operating characteristics (AUC_ROC) through a ten-fold cross validation. Results We analyzed data obtained from 17,658 inpatients with diabetes who underwent 32,758 admissions between July 2014 and August 2018. The predictive factors from the logistic regression model included people undergoing procedures, weight, type of diabetes, oxygen saturation level, use of medications (insulin, sulfonylurea, metformin) and albumin levels. The machine learning model with the best performance was the XGBoost model (AUC_ROC 0.96. This outperformed the logistic regression model which had an AUC_ROC of 0.75 for the estimation of the risk of clinically significant hypoglycaemia. Conclusions Advanced machine learning models are superior to logistic regression models in predicting the risk of hypoglycemia in inpatients with diabetes. Trials of such models should be conducted in real time to evaluate their utility to reduce inpatient hypoglycaemia.

Download Full-text

Predicting the risk of inpatient hypoglycemia with machine learning using electronic health records

10.2337/figshare.12091953 ◽

2020 ◽

Author(s):

Yue Ruan ◽

Alexis Bellot ◽

Zuzana Moysova ◽

Garry D. Tan ◽

Alistair Lumb ◽

...

Keyword(s):

Machine Learning ◽

Logistic Regression ◽

Regression Model ◽

Logistic Regression Model ◽

Prediction Models ◽

Machine Learning Algorithms ◽

Operating Characteristics ◽

Oxygen Saturation Level ◽

Electronic Health ◽

Clinically Significant

Download Full-text

A machine learning-based prediction of hospital mortality in patients with postoperative sepsis

10.21203/rs.2.24188/v1 ◽

2020 ◽

Cited By ~ 1

Author(s):

Ren-qi Yao ◽

Xin Jin ◽

Guo-wei Wang ◽

Yue Yu ◽

Guo-sheng Wu ◽

...

Keyword(s):

Machine Learning ◽

Logistic Regression ◽

Hospital Mortality ◽

Regression Model ◽

Logistic Regression Model ◽

Gradient Boosting ◽

Stepwise Logistic Regression ◽

Postoperative Sepsis ◽

Extreme Gradient Boosting ◽

Stepwise Logistic Regression Model

Abstract Background: The incidence of postoperative sepsis is continually increased, while few studies have specifically focused on the risk factors and clinical outcomes associated with the development of sepsis after surgical procedures. The present study aimed to develop a mathematical model for predicting the in-hospital mortality among patients with postoperative sepsis.Methods: Surgical patients in Medical Information Mart for Intensive Care (MIMIC-III) database who simultaneously fulfilled Sepsis 3.0 as well as Agency for Healthcare Research and Quality (AHRQ) criteria during ICU admission were incorporated. We employed both extreme gradient boosting (XGBoost) and stepwise logistic regression model to predict in-hospital mortality among included patients with postoperative sepsis. Consequently, model performance was assessed from the angles of discrimination and calibration.Results: We included 3713 patients who fulfilled our inclusion criteria, in which 397 (10.7%) patients died during hospitalization, while 3316 (89.3%) of them survived through discharge. Fluid-electrolyte disturbance, coagulopathy, renal replacement therapy (RRT), urine output, and cardiovascular surgery were important features related to the in-hospital mortality. The XGBoost model had a better performance in both discriminatory ability (c-statistics, 0.835 [95% CI, 0.786 to 0.877] vs. c-statistics, 0.737 [95% CI, 0.688 to 0.786]) and goodness of fit (visualized by calibration curve) compared to the stepwise logistic regression model. Conclusion: XGBoost model appears to be a better performance in predicting hospital mortality among postoperative septic patients compared to the conventional stepwise logistic regression model. Machine learning-based algorithm might have significant application in the development of early warning system for septic patients following major operations.

Download Full-text

Effect of fluid administration on scene to traffic accident patients by EMS personnel: a propensity score-matched study using population-based ambulance records and nationwide trauma registry in Japan

European Journal of Trauma and Emergency Surgery ◽

10.1007/s00068-020-01590-z ◽

2021 ◽

Author(s):

Yusuke Katayama ◽

Tetsuhisa Kitamura ◽

Kosuke Kiyohara ◽

Kenichiro Ishida ◽

Tomoya Hirose ◽

...

Keyword(s):

Logistic Regression ◽

Propensity Score ◽

Hospital Mortality ◽

Regression Model ◽

Traffic Accident ◽

Logistic Regression Model ◽

Cardiopulmonary Arrest ◽

Population Based ◽

Fluid Administration ◽

Secondary Outcome

Abstract Purpose The aim of this study was to assess the effect of fluid administration by emergency life-saving technicians (ELST) on the prognosis of traffic accident patients by using a propensity score (PS)-matching method. Methods The study included traffic accident patients registered in the JTDB database from January 2016 to December 2017. The main outcome was hospital mortality, and the secondary outcome was cardiopulmonary arrest on hospital arrival (CPAOA). To reduce potential confounding effects in the comparisons between two groups, we estimated a propensity score (PS) by fitting a logistic regression model that was adjusted for 17 variables before the implementation of fluid administration by ELST at the scene. Results During the study period, 10,908 traffic accident patients were registered in the JTDB database, and we included 3502 patients in this study. Of these patients, 142 were administered fluid by ELST and 3360 were not administered fluid by ELST. After PS matching, 141 patients were selected from each group. In the PS-matched model, fluid administration by ELST at the scene was not associated with discharge to death (crude OR: 0.859 [95% CI, 0.500–1.475]; p = 0.582). However, the fluid group showed statistically better outcome for CPAOA than the no fluid group in the multiple logistic regression model (adjusted OR: 0.231 [95% CI, 0.055–0.967]; p = 0.045). Conclusion In this study, fluid administration to traffic accident patients by ELST was associated not with hospital mortality but with a lower proportion of CPAOA.

Download Full-text

A comprehensive, contemporary assessment of the association between hepatosteatosis and coronary artery calcium scoring

European Heart Journal - Cardiovascular Imaging ◽

10.1093/ehjci/jeaa356.234 ◽

2021 ◽

Vol 22 (Supplement_1) ◽

Author(s):

T Heseltine ◽

SW Murray ◽

RL Jones ◽

M Fisher ◽

B Ruzsics

Keyword(s):

Risk Factors ◽

Logistic Regression ◽

Coronary Artery ◽

Regression Model ◽

Coronary Artery Calcium ◽

Logistic Regression Model ◽

Prediction Models ◽

Multinomial Logistic Regression ◽

Cvd Risk ◽

Male Sex

Abstract Funding Acknowledgements Type of funding sources: None. onbehalf Liverpool Multiparametric Imaging Collaboration Background Coronary artery calcium (CAC) score is a well-established technique for stratifying an individual’s cardiovascular disease (CVD) risk. Several well-established registries have incorporated CAC scoring into CVD risk prediction models to enhance accuracy. Hepatosteatosis (HS) has been shown to be an independent predictor of CVD events and can be measured on non-contrast computed tomography (CT). We sought to undertake a contemporary, comprehensive assessment of the influence of HS on CAC score alongside traditional CVD risk factors. In patients with HS it may be beneficial to offer routine CAC screening to evaluate CVD risk to enhance opportunities for earlier primary prevention strategies. Methods We performed a retrospective, observational analysis at a high-volume cardiac CT centre analysing consecutive CT coronary angiography (CTCA) studies. All patients referred for investigation of chest pain over a 28-month period (June 2014 to November 2016) were included. Patients with established CVD were excluded. The cardiac findings were reported by a cardiologist and retrospectively analysed by two independent radiologists for the presence of HS. Those with CAC of zero and those with CAC greater than zero were compared for demographic and cardiac risks. A multivariate analysis comparing the risk factors was performed to adjust for the presence of established risk factors. A binomial logistic regression model was developed to assess the association between the presence of HS and increasing strata of CAC. Results In total there were 1499 patients referred for CTCA without prior evidence of CVD. The assessment of HS was completed in 1195 (79.7%) and CAC score was performed in 1103 (92.3%). There were 466 with CVD and 637 without CVD. The prevalence of HS was significantly higher in those with CVD versus those without CVD on CTCA (51.3% versus 39.9%, p = 0.007). Male sex (50.7% versus 36.1% p= <0.001), age (59.4 ± 13.7 versus 48.1 ± 13.6, p= <0.001) and diabetes (12.4% versus 6.9%, p = 0.04) were also significantly higher in the CAC group compared to the CAC score of zero. HS was associated with increasing strata of CAC score compared with CAC of zero (CAC score 1-100 OR1.47, p = 0.01, CAC score 101-400 OR:1.68, p = 0.02, CAC score >400 OR 1.42, p = 0.14). This association became non-significant in the highest strata of CAC score. Conclusion We found a significant association between the increasing age, male sex, diabetes and HS with the presence of CAC. HS was also associated with a more severe phenotype of CVD based on the multinomial logistic regression model. Although the association reduced for the highest strata of CAC (CAC score >400) this likely reflects the overall low numbers of patients within this group and is likely a type II error. Based on these findings it may be appropriate to offer routine CVD risk stratification techniques in all those diagnosed with HS.

Download Full-text

Logistic Regression Model for Loan Prediction: A Machine Learning Approach

10.1109/eti4.051663.2021.9619201 ◽

2021 ◽

Author(s):

Richa Manglani ◽

Anuja Bokhare

Keyword(s):

Machine Learning ◽

Logistic Regression ◽

Regression Model ◽

Logistic Regression Model ◽

Learning Approach ◽

Machine Learning Approach

Download Full-text

Machine Learning-based in-hospital Mortality Prediction Models for Patients With Acute Coronary Syndrome

10.21203/rs.3.rs-134944/v1 ◽

2020 ◽

Author(s):

Jun Ke ◽

Yiwei Chen ◽

Xiaoping Wang ◽

Zhiyong Wu ◽

qiongyao Zhang ◽

...

Keyword(s):

Machine Learning ◽

Logistic Regression ◽

Random Forest ◽

Hospital Mortality ◽

Operating Characteristic ◽

Prediction Models ◽

Characteristic Curve ◽

Multivariate Logistic Regression Analysis ◽

Hdl Cholesterol ◽

Coronary Syndrome

Abstract BackgroundThe purpose of this study is to identify the risk factors of in-hospital mortality in patients with acute coronary syndrome (ACS) and to evaluate the performance of traditional regression and machine learning prediction models.MethodsThe data of ACS patients who entered the emergency department of Fujian Provincial Hospital from January 1, 2017 to March 31, 2020 for chest pain were retrospectively collected. The study used univariate and multivariate logistic regression analysis to identify risk factors for in-hospital mortality of ACS patients. The traditional regression and machine learning algorithms were used to develop predictive models, and the sensitivity, specificity, and receiver operating characteristic curve were used to evaluate the performance of each model.ResultsA total of 7810 ACS patients were included in the study, and the in-hospital mortality rate was 1.75%. Multivariate logistic regression analysis found that age and levels of D-dimer, cardiac troponin I, N-terminal pro-B-type natriuretic peptide (NT-proBNP), lactate dehydrogenase (LDH), high-density lipoprotein (HDL) cholesterol, and calcium channel blockers were independent predictors of in-hospital mortality. The study found that the area under the receiver operating characteristic curve of the models developed by logistic regression, gradient boosting decision tree (GBDT), random forest, and support vector machine (SVM) for predicting the risk of in-hospital mortality were 0.963, 0.960, 0.963, and 0.959, respectively. Feature importance evaluation found that NT-proBNP, LDH, and HDL cholesterol were top three variables that contribute the most to the prediction performance of the GBDT model and random forest model.ConclusionsThe predictive model developed using logistic regression, GBDT, random forest, and SVM algorithms can be used to predict the risk of in-hospital death of ACS patients. Based on our findings, we recommend that clinicians focus on monitoring the changes of NT-proBNP, LDH, and HDL cholesterol, as this may improve the clinical outcomes of ACS patients.

Download Full-text

Machine Learning for Mortality Prediction in Pediatric Myocarditis

Frontiers in Pediatrics ◽

10.3389/fped.2021.644922 ◽

2021 ◽

Vol 9 ◽

Author(s):

Fu-Sheng Chou ◽

Laxmi V. Ghimire

Keyword(s):

Machine Learning ◽

Linear Regression ◽

Regression Model ◽

Current Knowledge ◽

External Validation ◽

Model Development ◽

Kidney Injury ◽

High Specificity ◽

Mortality Prediction ◽

Linear Regression Models

Background: Pediatric myocarditis is a rare disease. The etiologies are multiple. Mortality associated with the disease is 5–8%. Prognostic factors were identified with the use of national hospitalization databases. Applying these identified risk factors for mortality prediction has not been reported.Methods: We used the Kids' Inpatient Database for this project. We manually curated fourteen variables as predictors of mortality based on the current knowledge of the disease, and compared performance of mortality prediction between linear regression models and a machine learning (ML) model. For ML, the random forest algorithm was chosen because of the categorical nature of the variables. Based on variable importance scores, a reduced model was also developed for comparison.Results: We identified 4,144 patients from the database for randomization into the primary (for model development) and testing (for external validation) datasets. We found that the conventional logistic regression model had low sensitivity (~50%) despite high specificity (>95%) or overall accuracy. On the other hand, the ML model struck a good balance between sensitivity (89.9%) and specificity (85.8%). The reduced ML model with top five variables (mechanical ventilation, cardiac arrest, ECMO, acute kidney injury, ventricular fibrillation) were sufficient to approximate the prediction performance of the full model.Conclusions: The ML algorithm performs superiorly when compared to the linear regression model for mortality prediction in pediatric myocarditis in this retrospective dataset. Prospective studies are warranted to further validate the applicability of our model in clinical settings.

Download Full-text

Early Plasma Osmolality Levels and Clinical Outcomes in Children Admitted to the Pediatric Intensive Care Unit: A Single-Center Cohort Study

Frontiers in Pediatrics ◽

10.3389/fped.2021.745204 ◽

2021 ◽

Vol 9 ◽

Author(s):

Huabin Wang ◽

Zhongyuan He ◽

Jiahong Li ◽

Chao Lin ◽

Huan Li ◽

...

Keyword(s):

Logistic Regression ◽

Intensive Care ◽

Hospital Mortality ◽

Regression Model ◽

Clinical Outcomes ◽

Poor Prognosis ◽

Logistic Regression Model ◽

Plasma Osmolality ◽

Pediatric Intensive Care ◽

Linear Spline

Objective: Identifying high-risk children with a poor prognosis in pediatric intensive care units (PICUs) is critical. The aim of this study was to assess the predictive value of early plasma osmolality levels in determining the clinical outcomes of children in PICUs.Methods: We retrospectively assessed critically ill children in a pediatric intensive care database. The locally weighted-regression scatter-plot smoothing (LOWESS) method was used to explore the approximate relationship between plasma osmolality and in-hospital mortality. Linear spline functions and stepwise expansion models were applied in conjunction with a multivariate logistic regression to further analyze this relationship. A subgroup analysis by age and complications was performed.Results: In total, 5,620 pediatric patients were included in this study. An approximately “U”-shaped relationship between plasma osmolality and mortality was detected using LOWESS. In the logistic regression model using a linear spline function, plasma osmolality ≥ 290 mmol/L was significantly associated with in-hospital mortality [odds ratio (OR) 1.020, 95% confidence interval (CI) 1.010–1.031], while plasma osmolality <290 mmol/L was not significantly associated with in-hospital mortality (OR 0.990, 95% CI 0.966–1.014). In the logistic regression model with plasma osmolality as a tri-categorical variable, only high osmolality was significantly associated with in-hospital mortality (OR 1.90, 95% CI 1.38–2.64), whereas low osmolality was not associated with in-hospital mortality (OR 1.28, 95% CI 0.84–1.94). The interactions between plasma osmolality and age or complications were not significant.Conclusion: High osmolality, rather than low osmolality, can predict a poor prognosis in children in PICUs.

Download Full-text

Machine-Learning vs. Expert-Opinion Driven Logistic Regression Modelling for Predicting 30-Day Unplanned Rehospitalisation in Preterm Babies: A Prospective, Population-Based Study (EPIPAGE 2)

Frontiers in Pediatrics ◽

10.3389/fped.2020.585868 ◽

2021 ◽

Vol 8 ◽

Author(s):

Robert A. Reed ◽

Andrei S. Morgan ◽

Jennifer Zeitlin ◽

Pierre-Henri Jarreau ◽

Héloïse Torchin ◽

...

Keyword(s):

Machine Learning ◽

Logistic Regression ◽

Random Forest ◽

Regression Model ◽

Expert Opinion ◽

Logistic Regression Model ◽

Population Based ◽

Regression Modelling ◽

Preterm Babies ◽

Logistic Regression Modelling

Introduction: Preterm babies are a vulnerable population that experience significant short and long-term morbidity. Rehospitalisations constitute an important, potentially modifiable adverse event in this population. Improving the ability of clinicians to identify those patients at the greatest risk of rehospitalisation has the potential to improve outcomes and reduce costs. Machine-learning algorithms can provide potentially advantageous methods of prediction compared to conventional approaches like logistic regression.Objective: To compare two machine-learning methods (least absolute shrinkage and selection operator (LASSO) and random forest) to expert-opinion driven logistic regression modelling for predicting unplanned rehospitalisation within 30 days in a large French cohort of preterm babies.Design, Setting and Participants: This study used data derived exclusively from the population-based prospective cohort study of French preterm babies, EPIPAGE 2. Only those babies discharged home alive and whose parents completed the 1-year survey were eligible for inclusion in our study. All predictive models used a binary outcome, denoting a baby's status for an unplanned rehospitalisation within 30 days of discharge. Predictors included those quantifying clinical, treatment, maternal and socio-demographic factors. The predictive abilities of models constructed using LASSO and random forest algorithms were compared with a traditional logistic regression model. The logistic regression model comprised 10 predictors, selected by expert clinicians, while the LASSO and random forest included 75 predictors. Performance measures were derived using 10-fold cross-validation. Performance was quantified using area under the receiver operator characteristic curve, sensitivity, specificity, Tjur's coefficient of determination and calibration measures.Results: The rate of 30-day unplanned rehospitalisation in the eligible population used to construct the models was 9.1% (95% CI 8.2–10.1) (350/3,841). The random forest model demonstrated both an improved AUROC (0.65; 95% CI 0.59–0.7; p = 0.03) and specificity vs. logistic regression (AUROC 0.57; 95% CI 0.51–0.62, p = 0.04). The LASSO performed similarly (AUROC 0.59; 95% CI 0.53–0.65; p = 0.68) to logistic regression.Conclusions: Compared to an expert-specified logistic regression model, random forest offered improved prediction of 30-day unplanned rehospitalisation in preterm babies. However, all models offered relatively low levels of predictive ability, regardless of modelling method.

Download Full-text