Development and Validation of a Machine-Learning Model for Prediction of Extubation Failure in Intensive Care Units

Development and external validation of a novel modality for rapid recognition of aortic dissection based on peripheral pulse oximetry waveforms

BACKGROUND

Aortic dissection (AD) is a life-threatening cardiovascular emergency that is often misdiagnosed as other chest pain conditions. Physiologically, AD may cause abnormalities in peripheral blood flow, which can be detected using pulse oximetry waveforms.

PURPOSE

This study aimed to assess the feasibility of identifying AD based on pulse oximetry waveforms and to highlight the key waveform features that play a crucial role in this diagnostic method.

METHODS

This prospective study employed high-risk chest pain cohorts from two emergency departments. The initial cohort was enriched with AD patients (n = 258, 47% AD) for model development, while the second cohort consisted of chest pain patients awaiting angiography (n = 71, 25% AD) and was used for external validation. Pulse oximetry waveforms from the four extremities were collected for each patient. After data preprocessing, a recognition model based on the random forest algorithm was trained using patients' gender, age, and waveform difference features extracted from the pulse oximetry waveforms. The performance of the model was evaluated using receiver operating characteristic (ROC) curve analysis and decision curve analysis (DCA). The importance of features was also assessed using Shapley Value and Gini importance.

RESULTS

The model demonstrated strong performance in identifying AD in both the training and external validation sets. In the training set, the model achieved an area under the ROC curve of 0.979 (95% CI: 0.961-0.990), sensitivity of 0.918 (95% CI: 0.873-0.955), specificity of 0.949 (95% CI: 0.912-0.985), and accuracy of 0.933 (95% CI: 0.904-0.959). In the external validation set, the model attained an area under the ROC curve of 0.855 (95% CI: 0.720-0.965), sensitivity of 0.889 (95% CI: 0.722-1.000), specificity of 0.698 (95% CI: 0.566-0.812), and accuracy of 0.794 (95% CI: 0.672-0.878). Decision curve analysis (DCA) further showed that the model provided a substantial net benefit for identifying AD. The median mean and median variance of the four limbs' signals were the most influential features in the recognition model.

CONCLUSIONS

This study demonstrated the feasibility and strong performance of identifying AD based on peripheral pulse oximetry waveforms in high-risk chest pain populations in the emergency setting. The findings also provided valuable insights for future human fluid dynamics simulations to elucidate the impact of AD on blood flow in greater detail.

Analysis of Factors and Clinical Outcomes of Planned Tracheal Extubation Failure in Neurosurgical Intensive Care Unit Patients

ABSTRACT

BACKGROUND: Planned tracheal extubation failure is a common occurrence among patients in the neurosurgical intensive care unit (NICU) because of the complex nature of neurocritical injuries, and the failure could result in a poor prognosis. METHODS: We observed and recorded the patients with tracheal intubation in the NICU of a hospital in Shanghai from June 2021 to December 2022 and analyzed data from planned tracheal extubation, categorizing patients by success or failure, and compared outcomes between the two groups while investigating contributing factors. RESULTS: A total of 156 patients were included, 133 (85.3%) of whom were successfully extubated and 23 (14.7%) were not. The results of logistic regression analysis demonstrated that the Glasgow Coma Scale score before extubation (OR, 0.643; 95% CI, 0.444-0.931; P = .020) and the frequency of respiratory secretions suctioning before tracheal extubation (OR, 0.098; 95% CI, 0.027-0.354; P < .001) were independent risk factors for extubation failure. We also found that the extubation failure group experienced a significantly longer ICU stay and incurred higher hospitalization costs. CONCLUSIONS: Poor Glasgow Coma Scale scores and a high frequency of respiratory secretions suctioning before tracheal extubation were the main factors contributing to tracheal extubation failure in NICU patients. To avoid tracheal extubation failure and adverse outcomes, these two factors should be carefully assessed before tracheal extubation.

Association between the anion-gap and 28-day mortality in critically ill adult patients with sepsis: A retrospective cohort study

Metabolic acidosis is usually associated with the severity of the condition of patients with sepsis or septic shock. Serum anion gap (AG) is one of the indicators of response metabolism. This study was performed to investigate whether the initial serum AG is associated with the 28-day mortality in critically ill adult patients with sepsis. This retrospective cohort study, a total of 15,047 patients with confirmed Sepsis disease from 2008 to 2019 from the Medical Information Mart for Intensive Care IV (MIMIC-IV) v1.0 database. The MIMIC-IV database is a comprehensive, de-identified clinical dataset originating from the Beth Israel Deaconess Medical Center in Boston, it includes extensive data on intensive care unit (ICU) patients, such as vital signs, lab results, and medication orders, spanning multiple years, accessible to researchers through an application process. AG can be obtained by direct extraction in the MIMIC-IV database (itemid = 50,868 from the laboratory events table of mimic_hosp), inclusion of AG values for the first test on first day of ICU admission. The patients were grouped into quartiles according to the AG interquartile range. The primary outcome was the 28-day mortality. Multiple logistic regression analysis was used to calculate the odds ratio (OR), while accounting for potential confounders, and the robustness of the results were evaluated in subgroup analyses. Among the 15,047 patients included in this study, the average age was 65.9 ± 16.0 years, 42.5% were female, 66.1% were Caucasian, and the 28-day mortality rate was 17.9% (2686/15,047). Multiple logistic regression analysis revealed the 28-day mortality in every increase of AG (per SD mEq/L), there is an associated 1.2 times (OR 1.2, 95% CI 1.12-1.29, P < .001) increase. Increased 28-day mortality (OR 1.53, 95% confidence interval 1.29-1.81, P < .001) in the group with the AG (15-18 mEq/L), and (OR 1.69, 95% confidence interval 1.4-2.04, P < .001) in the group with the highest AG (≥18 mEq/L), AG (<12 mEq/L) as a reference group, in the fully adjusted model. In adult patients with sepsis, the early AG at the time of ICU admission is an independent risk factor for prognosis.

Risk Factors and Prediction of 28-Day-All Cause Mortality Among Critically Ill Patients with Acute Pancreatitis Using Machine Learning Techniques: A Retrospective Analysis of Multi-Institutions

Objective

This study aimed to identify the risk factors and construct a reliable prediction model of 28-day all-cause mortality in critically ill patients with acute pancreatitis (AP) using machine learning techniques.

Methods

A total of 534 patients from three different institutions were included. Thirty-eight possible variables were collected from the Intensive care unit (ICU) admission for investigation. Patients were split into a training cohort (n = 400) and test cohort (n = 134) according to their source of hospital. The synthetic minority oversampling technique (SMOTE) was introduced to handle the inherent class imbalance. Six machine learning algorithms were applied in this study. The optimal machine learning model was chosen after patients in the test cohort were selected to validate the models. SHapley Additive exPlanation (SHAP) analysis was performed to rank the importance of variable. The predictive performance of the models was evaluated by the calibration curve, area under the receiver operating characteristics curves (AUROC), and decision clinical analysis.

Results

About 13.5% (72/534) of all patients eventually died of all-cause within 28 days of ICU admission. Eight important variables were screened out, including white blood cell count, platelets, body temperature, age, blood urea nitrogen, red blood cell distribution width, SpO2, and hemoglobin. The support vector machine (SVM) algorithm performed best in predicting 28-d all-cause death. Its AUROC reached 0.877 (95% CI: 0.809 to 0.927, p < 0.001), the Youden index was 0.634 (95% CI: 0.459 to 0.717). Based on the risk stratification system, the difference between the high-risk and low-risk groups was significantly different.

Conclusion

In conclusion, this study developed and validated SVM model, which better predicted 28-d all-cause mortality in critically ill patients with AP. In the future, we will continue to include patients from more institutions to conduct validation in different contexts and countries.

Independent Risk Factors of Failed Extubation among Adult Critically Ill Patients: A Prospective Observational Study from Saudi Arabia

Background

Mechanical ventilation provides essential support for critically ill patients in several diagnoses; however, extubation failure can affect patient outcomes. From Saudi Arabia, no study has assessed the factors associated with extubation failure in adults.

Methods

This prospective observational study was conducted in the intensive care unit of a tertiary care hospital in Riyadh, Saudi Arabia. Adult patients who had been mechanically ventilated via the endotracheal tube for a minimum of 24 hours and then extubated according to the weaning protocol were included. Failed extubation was defined as reintubation within 48 hours of extubation.

Results

A total of 505 patients were included, of which 72 patients had failed extubation (14.3%, 95% CI: 11.4%-17.7%). Compared with the failed extubation group, the successfully extubated group had significantly shorter duration of mechanical ventilation (mean difference: -2.6 days, 95% CI: -4.3 to -1; P = 0.001), a slower respiratory rate at the time of extubation (mean difference: -2.3 breath/min, 95% CI: -3.8 to -1; P = 0.0005), higher pH (mean difference: 0.02, 95% CI: 0.001-0.04; P = 0.03), and more patients with strong cough (percent difference: 17.7%, 95% CI: 4.8%-30.5%; P = 0.02). Independent risk factors of failed extubation were age (aOR = 1.02; 95% CI: 1.002-1.03; P = 0.03), respiratory rate (aOR = 1.06, 95% CI: 1.01-1.1; P = 0.008), duration of mechanical ventilation (aOR = 1.08, 95% CI: 1.03 - 1.1; P < 0.001), and pH (aOR = 0.02, 95% CI: 0.0006-0.5; P = 0.02).

Conclusion

Older age, longer duration of mechanical ventilation, faster respiratory rate, and lower pH were found to be independent risk factors that significantly increased the odds of extubation failure among adults.

The intervention of artificial intelligence to improve the weaning outcomes of patients with mechanical ventilation: Practical applications in the medical intensive care unit and the COVID-19 intensive care unit: A retrospective study

Patients admitted to intensive care units (ICU) and receiving mechanical ventilation (MV) may experience ventilator-associated adverse events and have prolonged ICU length of stay (LOS). We conducted a survey on adult patients in the medical ICU requiring MV. Utilizing big data and artificial intelligence (AI)/machine learning, we developed a predictive model to determine the optimal timing for weaning success, defined as no reintubation within 48 hours. An interdisciplinary team integrated AI into our MV weaning protocol. The study was divided into 2 parts. The first part compared outcomes before AI (May 1 to Nov 30, 2019) and after AI (May 1 to Nov 30, 2020) implementation in the medical ICU. The second part took place during the COVID-19 pandemic, where patients were divided into control (without AI assistance) and intervention (with AI assistance) groups from Aug 1, 2022, to Apr 30, 2023, and we compared their short-term outcomes. In the first part of the study, the intervention group (with AI, n = 1107) showed a shorter mean MV time (144.3 hours vs 158.7 hours, P = .077), ICU LOS (8.3 days vs 8.8 days, P = .194), and hospital LOS (22.2 days vs 25.7 days, P = .001) compared to the pre-intervention group (without AI, n = 1298). In the second part of the study, the intervention group (with AI, n = 88) exhibited a shorter mean MV time (244.2 hours vs 426.0 hours, P = .011), ICU LOS (11.0 days vs 18.7 days, P = .001), and hospital LOS (23.5 days vs 40.4 days, P < .001) compared to the control group (without AI, n = 43). The integration of AI into the weaning protocol led to improvements in the quality and outcomes of MV patients.

Using Artificial Intelligence to Predict Mechanical Ventilation Weaning Success in Patients with Respiratory Failure, Including Those with Acute Respiratory Distress Syndrome

The management of mechanical ventilation (MV) remains a challenge in intensive care units (ICUs). The digitalization of healthcare and the implementation of artificial intelligence (AI) and machine learning (ML) has significantly influenced medical decision-making capabilities, potentially enhancing patient outcomes. Acute respiratory distress syndrome, an overwhelming inflammatory lung disease, is common in ICUs. Most patients require MV. Prolonged MV is associated with an increased length of stay, morbidity, and mortality. Shortening the MV duration has both clinical and economic benefits and emphasizes the need for better MV weaning management. AI and ML models can assist the physician in weaning patients from MV by providing predictive tools based on big data. Many ML models have been developed in recent years, dealing with this unmet need. Such models provide an important prediction regarding the success of the individual patient's MV weaning. Some AI models have shown a notable impact on clinical outcomes. However, there are challenges in integrating AI models into clinical practice due to the unfamiliar nature of AI for many physicians and the complexity of some AI models. Our review explores the evolution of weaning methods up to and including AI and ML as weaning aids.

Development of a clinical prediction tool for extubation failure in pediatric cardiac intensive care unit

Introduction/objective

Extubation failure in pediatric patients with congenital or acquired heart diseases increases morbidity and mortality. This study aimed to develop a clinical risk score for predicting extubation failure to guide proper clinical decision-making and management.

Methods

We conducted a retrospective study. This clinical prediction score was developed using data from the Pediatric Cardiac Intensive Care Unit (PCICU) of the Faculty of Medicine, Chiang Mai University, Thailand, from July 2016 to May 2022. Extubation failure was defined as the requirement for re-intubation within 48 h after extubation. Multivariable logistic regression was used for modeling. The score was evaluated in terms of discrimination and calibration.

Results

A total of 352 extubation events from 270 patients were documented. Among these, 40 events (11.36%) were extubation failure. Factors associated with extubation failure included history of pneumonia (OR: 4.14, 95% CI: 1.83-9.37, p = 0.001), history of re-intubation (OR: 5.99, 95% CI: 2.12-16.98, p = 0.001), and high saturation in physiologic cyanosis (OR: 5.94, 95% CI: 1.87-18.84, p = 0.003). These three factors were utilized to develop the risk score. The score showed acceptable discrimination with an area under the curve (AUC) of 0.77 (95% CI: 0.69-0.86), and good calibration.

Conclusion

The derived Pediatric CMU Extubation Failure Prediction Score (Ped-CMU ExFPS) could satisfactorily predict extubation failure in pediatric cardiac patients. Employing this score could promote proper personalized care. We suggest conducting further external validation studies before considering implementation in practice.

The prediction of in-hospital mortality in elderly patients with sepsis-associated acute kidney injury utilizing machine learning models

Background

Sepsis-associated acute kidney injury (SA-AKI) is a severe complication associated with poorer prognosis and increased mortality, particularly in elderly patients. Currently, there is a lack of accurate mortality risk prediction models for these patients in clinic.

Objectives

This study aimed to develop and validate machine learning models for predicting in-hospital mortality risk in elderly patients with SA-AKI.

Methods

Machine learning models were developed and validated using the public, high-quality Medical Information Mart for Intensive Care (MIMIC)-IV critically ill database. The recursive feature elimination (RFE) algorithm was employed for key feature selection. Eleven predictive models were compared, with the best one selected for further validation. Shapley Additive Explanations (SHAP) values were used for visualization and interpretation, making the machine learning models clinically interpretable.

Results

There were 16,154 patients with SA-AKI in the MIMIC-IV database, and 8426 SA-AKI patients were included in this study (median age: 77.0 years; female: 45%). 7728 patients excluded based on these criteria. They were randomly divided into a training cohort (5,934, 70%) and a validation cohort (2,492, 30%). Nine key features were selected by the RFE algorithm. The CatBoost model achieved the best performance, with an AUC of 0.844 in the training cohort and 0.804 in the validation cohort. SHAP values revealed that AKI stage, PaO2, and lactate were the top three most important features contributing to the CatBoost model.

Conclusion

We developed a model capable of predicting the risk of in-hospital mortality in elderly patients with SA-AKI.

Predicting risk of preterm birth in singleton pregnancies using machine learning algorithms

We aimed to develop, train, and validate machine learning models for predicting preterm birth (<37 weeks' gestation) in singleton pregnancies at different gestational intervals. Models were developed based on complete data from 22,603 singleton pregnancies from a prospective population-based cohort study that was conducted in 51 midwifery clinics and hospitals in Wenzhou City of China between 2014 and 2016. We applied Catboost, Random Forest, Stacked Model, Deep Neural Networks (DNN), and Support Vector Machine (SVM) algorithms, as well as logistic regression, to conduct feature selection and predictive modeling. Feature selection was implemented based on permutation-based feature importance lists derived from the machine learning models including all features, using a balanced training data set. To develop prediction models, the top 10%, 25%, and 50% most important predictive features were selected. Prediction models were developed with the training data set with 5-fold cross-validation for internal validation. Model performance was assessed using area under the receiver operating curve (AUC) values. The CatBoost-based prediction model after 26 weeks' gestation performed best with an AUC value of 0.70 (0.67, 0.73), accuracy of 0.81, sensitivity of 0.47, and specificity of 0.83. Number of antenatal care visits before 24 weeks' gestation, aspartate aminotransferase level at registration, symphysis fundal height, maternal weight, abdominal circumference, and blood pressure emerged as strong predictors after 26 completed weeks. The application of machine learning on pregnancy surveillance data is a promising approach to predict preterm birth and we identified several modifiable antenatal predictors.

Predicting Hypoxia Using Machine Learning: Systematic Review

Background

Hypoxia is an important risk factor and indicator for the declining health of inpatients. Predicting future hypoxic events using machine learning is a prospective area of study to facilitate time-critical interventions to counter patient health deterioration.

Objective

This systematic review aims to summarize and compare previous efforts to predict hypoxic events in the hospital setting using machine learning with respect to their methodology, predictive performance, and assessed population.

Methods

A systematic literature search was performed using Web of Science, Ovid with Embase and MEDLINE, and Google Scholar. Studies that investigated hypoxia or hypoxemia of hospitalized patients using machine learning models were considered. Risk of bias was assessed using the Prediction Model Risk of Bias Assessment Tool.

Results

After screening, a total of 12 papers were eligible for analysis, from which 32 models were extracted. The included studies showed a variety of population, methodology, and outcome definition. Comparability was further limited due to unclear or high risk of bias for most studies (10/12, 83%). The overall predictive performance ranged from moderate to high. Based on classification metrics, deep learning models performed similar to or outperformed conventional machine learning models within the same studies. Models using only prior peripheral oxygen saturation as a clinical variable showed better performance than models based on multiple variables, with most of these studies (2/3, 67%) using a long short-term memory algorithm.

Conclusions

Machine learning models provide the potential to accurately predict the occurrence of hypoxic events based on retrospective data. The heterogeneity of the studies and limited generalizability of their results highlight the need for further validation studies to assess their predictive performance.

Development of a Machine Learning Model for Predicting Weaning Outcomes Based Solely on Continuous Ventilator Parameters during Spontaneous Breathing Trials

Discontinuing mechanical ventilation remains challenging. We developed a machine learning model to predict weaning outcomes using only continuous monitoring parameters obtained from ventilators during spontaneous breathing trials (SBTs). Patients who received mechanical ventilation in the medical intensive care unit at a tertiary university hospital from 2019-2021 were included in this study. During the SBTs, three waveforms and 25 numerical data were collected as input variables. The proposed convolutional neural network (CNN)-based weaning prediction model extracts features from input data with diverse lengths. Among 138 enrolled patients, 35 (25.4%) experienced weaning failure. The dataset was randomly divided into training and test sets (8:2 ratio). The area under the receiver operating characteristic curve for weaning success by the prediction model was 0.912 (95% confidence interval [CI], 0.795-1.000), with an area under the precision-recall curve of 0.767 (95% CI, 0.434-0.983). Furthermore, we used gradient-weighted class activation mapping technology to provide visual explanations of the model's prediction, highlighting influential features. This tool can assist medical staff by providing intuitive information regarding readiness for extubation without requiring any additional data collection other than SBT data. The proposed predictive model can assist clinicians in making ventilator weaning decisions in real time, thereby improving patient outcomes.

An interpretable deep learning framework for predicting liver metastases in postoperative colorectal cancer patients using natural language processing and clinical data integration

BACKGROUND

The significance of liver metastasis (LM) in increasing the risk of death for postoperative colorectal cancer (CRC) patients necessitates innovative approaches to predict LM.

AIM

Our study presents a novel and significant contribution by developing an interpretable fusion model that effectively integrates both free-text medical record data and structured laboratory data to predict LM in postoperative CRC patients.

METHODS

We used a robust dataset of 1463 patients and leveraged state-of-the-art natural language processing (NLP) and machine learning techniques to construct a two-layer fusion framework that demonstrates superior predictive performance compared to single modal models. Our innovative two-tier algorithm fuses the results from different data modalities, achieving balanced prediction results on test data and significantly enhancing the predictive ability of the model. To increase interpretability, we employed Shapley additive explanations to elucidate the contributions of free-text clinical data and structured clinical data to the final model. Furthermore, we translated our findings into practical clinical applications by creating a novel NLP score-based nomogram using the top 13 valid predictors identified in our study.

RESULTS

The proposed fusion models demonstrated superior predictive performance with an accuracy of 80.8%, precision of 80.3%, recall of 80.5%, and an F1 score of 80.8% in predicting LMs.

CONCLUSION

This fusion model represents a notable advancement in predicting LMs for postoperative CRC patients, offering the potential to enhance patient outcomes and support clinical decision-making.

Risk factor mining and prediction of urine protein progression in chronic kidney disease: a machine learning- based study

BACKGROUND

Chronic kidney disease (CKD) is a global public health concern. Therefore, to provide timely intervention for non-hospitalized high-risk patients and rationally allocate limited clinical resources is important to mine the key factors when designing a CKD prediction model.

METHODS

This study included data from 1,358 patients with CKD pathologically confirmed during the period from December 2017 to September 2020 at Zhongshan Hospital. A CKD prediction interpretation framework based on machine learning was proposed. From among 100 variables, 17 were selected for the model construction through a recursive feature elimination with logistic regression feature screening. Several machine learning classifiers, including extreme gradient boosting, gaussian-based naive bayes, a neural network, ridge regression, and linear model logistic regression (LR), were trained, and an ensemble model was developed to predict 24-hour urine protein. The detailed relationship between the risk of CKD progression and these predictors was determined using a global interpretation. A patient-specific analysis was conducted using a local interpretation.

RESULTS

The results showed that LR achieved the best performance, with an area under the curve (AUC) of 0.850 in a single machine learning model. The ensemble model constructed using the voting integration method further improved the AUC to 0.856. The major predictors of moderate-to-severe severity included lower levels of 25-OH-vitamin, albumin, transferrin in males, and higher levels of cystatin C.

CONCLUSIONS

Compared with the clinical single kidney function evaluation indicators (eGFR, Scr), the machine learning model proposed in this study improved the prediction accuracy of CKD progression by 17.6% and 24.6%, respectively, and the AUC was improved by 0.250 and 0.236, respectively. Our framework can achieve a good predictive interpretation and provide effective clinical decision support.

Interpretable machine learning for predicting 28-day all-cause in-hospital mortality for hypertensive ischemic or hemorrhagic stroke patients in the ICU: a multi-center retrospective cohort study with internal and external cross-validation

Background

Timely and accurate outcome prediction plays a critical role in guiding clinical decisions for hypertensive ischemic or hemorrhagic stroke patients admitted to the ICU. However, interpreting and translating the predictive models into clinical applications are as important as the prediction itself. This study aimed to develop an interpretable machine learning (IML) model that accurately predicts 28-day all-cause mortality in hypertensive ischemic or hemorrhagic stroke patients.

Methods

A total of 4,274 hypertensive ischemic or hemorrhagic stroke patients admitted to the ICU in the USA from multicenter cohorts were included in this study to develop and validate the IML model. Five machine learning (ML) models were developed, including artificial neural network (ANN), gradient boosting machine (GBM), eXtreme Gradient Boosting (XGBoost), logistic regression (LR), and support vector machine (SVM), to predict mortality using the MIMIC-IV and eICU-CRD database in the USA. Feature selection was performed using the Least Absolute Shrinkage and Selection Operator (LASSO) algorithm. Model performance was evaluated based on the area under the curve (AUC), accuracy, positive predictive value (PPV), and negative predictive value (NPV). The ML model with the best predictive performance was selected for interpretability analysis. Finally, the SHapley Additive exPlanations (SHAP) method was employed to evaluate the risk of all-cause in-hospital mortality among hypertensive ischemic or hemorrhagic stroke patients admitted to the ICU.

Results

The XGBoost model demonstrated the best predictive performance, with the AUC values of 0.822, 0.739, and 0.700 in the training, test, and external cohorts, respectively. The analysis of feature importance revealed that age, ethnicity, white blood cell (WBC), hyperlipidemia, mean corpuscular volume (MCV), glucose, pulse oximeter oxygen saturation (SpO2), serum calcium, red blood cell distribution width (RDW), blood urea nitrogen (BUN), and bicarbonate were the 11 most important features. The SHAP plots were employed to interpret the XGBoost model.

Conclusions

The XGBoost model accurately predicted 28-day all-cause in-hospital mortality among hypertensive ischemic or hemorrhagic stroke patients admitted to the ICU. The SHAP method can provide explicit explanations of personalized risk prediction, which can aid physicians in understanding the model.

Developing a machine-learning model for real-time prediction of successful extubation in mechanically ventilated patients using time-series ventilator-derived parameters

Background

Successful weaning from mechanical ventilation is important for patients admitted to intensive care units. However, models for predicting real-time weaning outcomes remain inadequate. Therefore, this study aimed to develop a machine-learning model for predicting successful extubation only using time-series ventilator-derived parameters with good accuracy.

Methods

Patients with mechanical ventilation admitted to the Yuanlin Christian Hospital in Taiwan between August 2015 and November 2020 were retrospectively included. A dataset with ventilator-derived parameters was obtained before extubation. Recursive feature elimination was applied to select the most important features. Machine-learning models of logistic regression, random forest (RF), and support vector machine were adopted to predict extubation outcomes. In addition, the synthetic minority oversampling technique (SMOTE) was employed to address the data imbalance problem. The area under the receiver operating characteristic (AUC), F1 score, and accuracy, along with the 10-fold cross-validation, were used to evaluate prediction performance.

Results

In this study, 233 patients were included, of whom 28 (12.0%) failed extubation. The six ventilatory variables per 180 s dataset had optimal feature importance. RF exhibited better performance than the others, with an AUC value of 0.976 (95% confidence interval [CI], 0.975-0.976), accuracy of 94.0% (95% CI, 93.8-94.3%), and an F1 score of 95.8% (95% CI, 95.7-96.0%). The difference in performance between the RF and the original and SMOTE datasets was small.

Conclusion

The RF model demonstrated a good performance in predicting successful extubation in mechanically ventilated patients. This algorithm made a precise real-time extubation outcome prediction for patients at different time points.

Deep learning approach for early prediction of COVID-19 mortality using chest X-ray and electronic health records

BACKGROUND

An artificial-intelligence (AI) model for predicting the prognosis or mortality of coronavirus disease 2019 (COVID-19) patients will allow efficient allocation of limited medical resources. We developed an early mortality prediction ensemble model for COVID-19 using AI models with initial chest X-ray and electronic health record (EHR) data.

RESULTS

We used convolutional neural network (CNN) models (Inception-ResNet-V2 and EfficientNet) for chest X-ray analysis and multilayer perceptron (MLP), Extreme Gradient Boosting (XGBoost), and random forest (RF) models for EHR data analysis. The Gradient-weighted Class Activation Mapping and Shapley Additive Explanations (SHAP) methods were used to determine the effects of these features on COVID-19. We developed an ensemble model (Area under the receiver operating characteristic curve of 0.8698) using a soft voting method with weight differences for CNN, XGBoost, MLP, and RF models. To resolve the data imbalance, we conducted F1-score optimization by adjusting the cutoff values to optimize the model performance (F1 score of 0.77).

CONCLUSIONS

Our study is meaningful in that we developed an early mortality prediction model using only the initial chest X-ray and EHR data of COVID-19 patients. Early prediction of the clinical courses of patients is helpful for not only treatment but also bed management. Our results confirmed the performance improvement of the ensemble model achieved by combining AI models. Through the SHAP method, laboratory tests that indicate the factors affecting COVID-19 mortality were discovered, highlighting the importance of these tests in managing COVID-19 patients.

Artificial intelligence for clinical decision support for monitoring patients in cardiovascular ICUs: A systematic review

Background

Artificial intelligence (AI) and machine learning (ML) models continue to evolve the clinical decision support systems (CDSS). However, challenges arise when it comes to the integration of AI/ML into clinical scenarios. In this systematic review, we followed the Preferred Reporting Items for Systematic reviews and Meta-Analyses (PRISMA), the population, intervention, comparator, outcome, and study design (PICOS), and the medical AI life cycle guidelines to investigate studies and tools which address AI/ML-based approaches towards clinical decision support (CDS) for monitoring cardiovascular patients in intensive care units (ICUs). We further discuss recent advances, pitfalls, and future perspectives towards effective integration of AI into routine practices as were identified and elaborated over an extensive selection process for state-of-the-art manuscripts.

Methods

Studies with available English full text from PubMed and Google Scholar in the period from January 2018 to August 2022 were considered. The manuscripts were fetched through a combination of the search keywords including AI, ML, reinforcement learning (RL), deep learning, clinical decision support, and cardiovascular critical care and patients monitoring. The manuscripts were analyzed and filtered based on qualitative and quantitative criteria such as target population, proper study design, cross-validation, and risk of bias.

Results

More than 100 queries over two medical search engines and subjective literature research were developed which identified 89 studies. After extensive assessments of the studies both technically and medically, 21 studies were selected for the final qualitative assessment.

Discussion

Clinical time series and electronic health records (EHR) data were the most common input modalities, while methods such as gradient boosting, recurrent neural networks (RNNs) and RL were mostly used for the analysis. Seventy-five percent of the selected papers lacked validation against external datasets highlighting the generalizability issue. Also, interpretability of the AI decisions was identified as a central issue towards effective integration of AI in healthcare.

A prediction model for predicting the risk of acute respiratory distress syndrome in sepsis patients: a retrospective cohort study

BACKGROUND

The risk of death in sepsis patients with acute respiratory distress syndrome (ARDS) was as high as 20-50%. Few studies focused on the risk identification of ARDS among sepsis patients. This study aimed to develop and validate a nomogram to predict the ARDS risk in sepsis patients based on the Medical Information Mart for Intensive Care IV database.

METHODS

A total of 16,523 sepsis patients were included and randomly divided into the training and testing sets with a ratio of 7:3 in this retrospective cohort study. The outcomes were defined as the occurrence of ARDS for ICU patients with sepsis. Univariate and multivariate logistic regression analyses were used in the training set to identify the factors that were associated with ARDS risk, which were adopted to establish the nomogram. The receiver operating characteristic and calibration curves were used to assess the predictive performance of nomogram.

RESULTS

Totally 2422 (20.66%) sepsis patients occurred ARDS, with the median follow-up time of 8.47 (5.20, 16.20) days. The results found that body mass index, respiratory rate, urine output, partial pressure of carbon dioxide, blood urea nitrogen, vasopressin, continuous renal replacement therapy, ventilation status, chronic pulmonary disease, malignant cancer, liver disease, septic shock and pancreatitis might be predictors. The area under the curve of developed model were 0.811 (95% CI 0.802-0.820) in the training set and 0.812 (95% CI 0.798-0.826) in the testing set. The calibration curve showed a good concordance between the predicted and observed ARDS among sepsis patients.

CONCLUSION

We developed a model incorporating thirteen clinical features to predict the ARDS risk in patients with sepsis. The model showed a good predictive ability by internal validation.

Identification of Distinct Clinical Phenotypes of Heterogeneous Mechanically Ventilated ICU Patients Using Cluster Analysis

This retrospective study aimed to derive the clinical phenotypes of ventilated ICU patients to predict the outcomes on the first day of ventilation. Clinical phenotypes were derived from the eICU Collaborative Research Database (eICU) cohort via cluster analysis and were validated in the Medical Information Mart for Intensive Care (MIMIC-IV) cohort. Four clinical phenotypes were identified and compared in the eICU cohort (n = 15,256). Phenotype A (n = 3112) was associated with respiratory disease, had the lowest 28-day mortality (16%), and had a high extubation success rate (~80%). Phenotype B (n = 3335) was correlated with cardiovascular disease, had the second-highest 28-day mortality (28%), and had the lowest extubation success rate (69%). Phenotype C (n = 3868) was correlated with renal dysfunction, had the highest 28-day mortality (28%), and had the second-lowest extubation success rate (74%). Phenotype D (n = 4941) was associated with neurological and traumatic diseases, had the second-lowest 28-day mortality (22%), and had the highest extubation success rate (>80%). These findings were validated in the validation cohort (n = 10,813). Additionally, these phenotypes responded differently to ventilation strategies in terms of duration of treatment, but had no difference in mortality. The four clinical phenotypes unveiled the heterogeneity of ICU patients and helped to predict the 28-day mortality and the extubation success rate.

Diagnostic efficacy of sonographic measurement of laryngeal air column width difference for predicting the risk of post-extubation stridor: A meta-analysis of observational studies

Background

This meta-analysis aimed at assessing the diagnostic accuracy of ultrasound-measured laryngeal air column width difference (ACWD) in predicting post-extubation stridor (PES) in intubated adult patients.

Methods

We searched the Medline, Cochrane Library, EMBASE, and Google scholar databases from inception to October, 2022 to identify studies that examined the diagnostic accuracy of ACWD for PES. The primary outcome was the diagnostic performance by calculating the pooled sensitivity, specificity, and area under the curve (AUC). The secondary outcomes were the differences in ACWD and duration of intubation between patients with and without PES.

Results

Following literature search, 11 prospective studies (intensive care setting, n = 10; operating room setting, n = 1) involving 1,322 extubations were included. The incidence of PES among the studies was 4-25%. All studies were mixed-gender (females: 24.1-68.5%) with sample sizes ranging between 41 and 432. The cut-off values of ACWD for prediction of PES varied from 0.45 to 1.6 mm. The pooled sensitivity and specificity of ACWD for PES were 0.8 (95% CI = 0.69-0.88, I ²: 37.26%, eight studies) and 0.81 (95% CI = 0.72-0.88, I ²: 89.51%, eight studies), respectively. The pooled AUC was 0.87 (95% CI = 0.84-0.90). Patients with PES had a smaller ACWD compared to those without PES (mean difference = -0.54, 95% CI = -0.79 to -0.28, I ²: 97%, eight studies). Moreover, patients with PES had a longer duration of tracheal intubation than that in those without (mean difference = 2.75 days, 95% CI = 0.92, 4.57, I ²: 90%, seven studies).

Conclusion

Ultrasound-measured laryngeal ACWD showed satisfactory sensitivity and specificity for predicting PES. Because of the limited number of studies available, further investigations are needed to support our findings.

Systematic review registration

https://www.crd.york.ac.uk/prospero/, identifier CRD42022375772.

Development and validation of a mechanical power-oriented prediction model of weaning failure in mechanically ventilated patients: a retrospective cohort study

OBJECTIVE

To develop and validate a mechanical power (MP)-oriented prediction model of weaning failure in mechanically ventilated patients.

DESIGN

A retrospective cohort study.

SETTING

Data were collected from the large US Medical Information Mart for Intensive Care-IV (MIMIC-IV) V.1.0, which integrates comprehensive clinical data from 76 540 intensive care unit (ICU) admissions from 2008 to 2019.

PARTICIPANTS

A total of 3695 patients with invasive mechanical ventilation for more than 24 hours and weaned with T-tube ventilation strategies were enrolled from the MIMIC-IV database.

PRIMARY AND SECONDARY OUTCOME

Weaning failure.

RESULTS

All eligible patients were randomised into development cohorts (n=2586, 70%) and validation cohorts (n=1109, 30%). Multivariate logistic regression analysis of the development cohort showed that positive end-expiratory pressure, dynamic lung compliance, MP, inspired oxygen concentration, length of ICU stay and invasive mechanical ventilation duration were independent predictors of weaning failure. Calibration curves showed good correlation between predicted and observed outcomes. The prediction model showed accurate discrimination in the development and validation cohorts, with area under the receiver operating characteristic curve values of 0.828 (95% CI: 0.812 to 0.844) and 0.833 (95% CI: 0.809 to 0.857), respectively. Decision curve analysis indicated that the predictive model was clinically beneficial.

CONCLUSION

The MP-oriented model of weaning failure accurately predicts the risk of weaning failure in mechanical ventilation patients and provides valuable information for clinicians making decisions on weaning.

Explainable machine learning approach to predict extubation in critically ill ventilated patients: a retrospective study in central Taiwan

Background

Weaning from mechanical ventilation (MV) is an essential issue in critically ill patients, and we used an explainable machine learning (ML) approach to establish an extubation prediction model.

Methods

We enrolled patients who were admitted to intensive care units during 2015–2019 at Taichung Veterans General Hospital, a referral hospital in central Taiwan. We used five ML models, including extreme gradient boosting (XGBoost), categorical boosting (CatBoost), light gradient boosting machine (LightGBM), random forest (RF) and logistic regression (LR), to establish the extubation prediction model, and the feature window as well as prediction window was 48 h and 24 h, respectively. We further employed feature importance, Shapley additive explanations (SHAP) plot, partial dependence plot (PDP) and local interpretable model-agnostic explanations (LIME) for interpretation of the model at the domain, feature, and individual levels.

Results

We enrolled 5,940 patients and found the accuracy was comparable among XGBoost, LightGBM, CatBoost and RF, with the area under the receiver operating characteristic curve using XGBoost to predict extubation was 0.921. The calibration and decision curve analysis showed well applicability of models. We also used the SHAP summary plot and PDP plot to demonstrate discriminative points of six key features in predicting extubation. Moreover, we employed LIME and SHAP force plots to show predicted probabilities of extubation and the rationale of the prediction at the individual level.

Conclusions

We developed an extubation prediction model with high accuracy and visualised explanations aligned with clinical workflow, and the model may serve as an autonomous screen tool for timely weaning.

Ensemble learning based on efficient features combination can predict the outcome of recurrence-free survival in patients with hepatocellular carcinoma within three years after surgery

Preoperative prediction of recurrence outcome in hepatocellular carcinoma (HCC) facilitates physicians' clinical decision-making. Preoperative imaging and related clinical baseline data of patients are valuable for evaluating prognosis. With the widespread application of machine learning techniques, the present study proposed the ensemble learning method based on efficient feature representations to predict recurrence outcomes within three years after surgery. Radiomics features during arterial phase (AP) and clinical data were selected for training the ensemble models. In order to improve the efficiency of the process, the lesion area was automatically segmented by 3D U-Net. It was found that the mIoU of the segmentation model was 0.8874, and the Light Gradient Boosting Machine (LightGBM) was the most superior, with an average accuracy of 0.7600, a recall of 0.7673, a F1 score of 0.7553, and an AUC of 0.8338 when inputting radiomics features during AP and clinical baseline indicators. Studies have shown that the proposed strategy can relatively accurately predict the recurrence outcome within three years, which is helpful for physicians to evaluate individual patients before surgery.

Interpretable recurrent neural network models for dynamic prediction of the extubation failure risk in patients with invasive mechanical ventilation in the intensive care unit

Background

Clinical decision of extubation is a challenge in the treatment of patient with invasive mechanical ventilation (IMV), since existing extubation protocols are not capable of precisely predicting extubation failure (EF). This study aims to develop and validate interpretable recurrent neural network (RNN) models for dynamically predicting EF risk.

Methods

A retrospective cohort study was conducted on IMV patients from the Medical Information Mart for Intensive Care IV (MIMIC-IV) database. Time series with a 4-h resolution were built for all included patients. Two types of RNN models, the long short-term memory (LSTM) and the gated recurrent unit (GRU), were developed. A stepwise logistic regression model was used to select key features for developing light-version RNN models. The RNN models were compared to other five non-temporal machine learning models. The Shapley additive explanations (SHAP) value was applied to explain the influence of the features on model prediction.

Results

Of 8,599 included patients, 2,609 had EF (30.3%). The area under receiver operating characteristic curve (AUROC) of LSTM and GRU showed no statistical difference on the test set (0.828 vs. 0.829). The light-version RNN models based on the 26 features selected out of a total of 89 features showed comparable performance as their corresponding full-version models. Among the non-temporal models, only the random forest (RF) (AUROC: 0.820) and the extreme gradient boosting (XGB) model (AUROC: 0.823) were comparable to the RNN models, but their calibration was deviated.

Conclusions

The RNN models have excellent predictive performance for predicting EF risk and have potential to become real-time assistant decision-making systems for extubation.

Supplementary Information

The online version contains supplementary material available at 10.1186/s13040-022-00309-7.

Data science in the intensive care unit

In this editorial, we comment on the current development and deployment of data science in intensive care units (ICUs). Data in ICUs can be classified into qualitative and quantitative data with different technologies needed to translate and interpret them. Data science, in the form of artificial intelligence (AI), should find the right interaction between physicians, data and algorithm. For individual patients and physicians, sepsis and mechanical ventilation have been two important aspects where AI has been extensively studied. However, major risks of bias, lack of generalizability and poor clinical values remain. AI deployment in the ICUs should be emphasized more to facilitate AI development. For ICU management, AI has a huge potential in transforming resource allocation. The coronavirus disease 2019 pandemic has given opportunities to establish such systems which should be investigated further. Ethical concerns must be addressed when designing such AI.

Machine learning for predicting successful extubation in patients receiving mechanical ventilation

Ventilator liberation is one of the most critical decisions in the intensive care unit; however, prediction of extubation failure is difficult, and the proportion thereof remains high. Machine learning can potentially provide a breakthrough in the prediction of extubation success. A total of seven studies on the prediction of extubation success using machine learning have been published. These machine learning models were developed using data from electronic health records, 8–78 features, and algorithms such as artificial neural network, LightGBM, and XGBoost. Sensitivity ranged from 0.64 to 0.96, specificity ranged from 0.73 to 0.85, and area under the receiver operating characteristic curve ranged from 0.70 to 0.98. The features deemed most important included duration of mechanical ventilation, PaO₂, blood urea nitrogen, heart rate, and Glasgow Coma Scale score. Although the studies had limitations, prediction of extubation success by machine learning has the potential to be a powerful tool. Further studies are needed to assess whether machine learning prediction reduces the incidence of extubation failure or prolongs the duration of ventilator use, thereby increasing tracheostomy and ventilator-related complications and mortality.

Early prediction of noninvasive ventilation failure after extubation: development and validation of a machine-learning model

Background

Noninvasive ventilation (NIV) has been widely used in critically ill patients after extubation. However, NIV failure is associated with poor outcomes. This study aimed to determine early predictors of NIV failure and to construct an accurate machine-learning model to identify patients at risks of NIV failure after extubation in intensive care units (ICUs).

Methods

Patients who underwent NIV after extubation in the eICU Collaborative Research Database (eICU-CRD) were included. NIV failure was defined as need for invasive ventilatory support (reintubation or tracheotomy) or death after NIV initiation. A total of 93 clinical and laboratory variables were assessed, and the recursive feature elimination algorithm was used to select key features. Hyperparameter optimization was conducted with an automated machine-learning toolkit called Neural Network Intelligence. A machine-learning model called Categorical Boosting (CatBoost) was developed and compared with nine other models. The model was then prospectively validated among patients enrolled in the Cardiac Surgical ICU of Zhongshan Hospital, Fudan University.

Results

Of 929 patients included in the eICU-CRD cohort, 248 (26.7%) had NIV failure. The time from extubation to NIV, age, Glasgow Coma Scale (GCS) score, heart rate, respiratory rate, mean blood pressure (MBP), saturation of pulse oxygen (SpO₂), temperature, glucose, pH, pressure of oxygen in blood (PaO₂), urine output, input volume, ventilation duration, and mean airway pressure were selected. After hyperparameter optimization, our model showed the greatest accuracy in predicting NIV failure (AUROC: 0.872 [95% CI 0.82–0.92]) among all predictive methods in an internal validation. In the prospective validation cohort, our model was also superior (AUROC: 0.846 [95% CI 0.80–0.89]). The sensitivity and specificity in the prediction group is 89% and 75%, while in the validation group they are 90% and 70%. MV duration and respiratory rate were the most important features. Additionally, we developed a web-based tool to help clinicians use our model.

Conclusions

This study developed and prospectively validated the CatBoost model, which can be used to identify patients who are at risk of NIV failure. Thus, those patients might benefit from early triage and more intensive monitoring.

Trial registration: NCT03704324. Registered 1 September 2018, https://register.clinicaltrials.gov.

Supplementary Information

The online version contains supplementary material available at 10.1186/s12890-022-02096-7.

Explainable machine learning models for predicting 30-day readmission in pediatric pulmonary hypertension: A multicenter, retrospective study

Background

Short-term readmission for pediatric pulmonary hypertension (PH) is associated with a substantial social and personal burden. However, tools to predict individualized readmission risk are lacking. This study aimed to develop machine learning models to predict 30-day unplanned readmission in children with PH.

Methods

This study collected data on pediatric inpatients with PH from the Chongqing Medical University Medical Data Platform from January 2012 to January 2019. Key clinical variables were selected by the least absolute shrinkage and the selection operator. Prediction models were selected from 15 machine learning algorithms with excellent performance, which was evaluated by area under the operating characteristic curve (AUC). The outcome of the predictive model was interpreted by SHapley Additive exPlanations (SHAP).

Results

A total of 5,913 pediatric patients with PH were included in the final cohort. The CatBoost model was selected as the predictive model with the greatest AUC for 0.81 (95% CI: 0.77–0.86), high accuracy for 0.74 (95% CI: 0.72–0.76), sensitivity 0.78 (95% CI: 0.69–0.87), and specificity 0.74 (95% CI: 0.72–0.76). Age, length of stay (LOS), congenital heart surgery, and nonmedical order discharge showed the greatest impact on 30-day readmission in pediatric PH, according to SHAP results.

Conclusions

This study developed a CatBoost model to predict the risk of unplanned 30-day readmission in pediatric patients with PH, which showed more significant performance compared with traditional logistic regression. We found that age, LOS, congenital heart surgery, and nonmedical order discharge were important factors for 30-day readmission in pediatric PH.

External validation based on transfer learning for diagnosing atelectasis using portable chest X-rays

Background

Although there has been a large amount of research focusing on medical image classification, few studies have focused specifically on the portable chest X-ray. To determine the feasibility of transfer learning method for detecting atelectasis with portable chest X-ray and its application to external validation, based on the analysis of a large dataset.

Methods

From the intensive care chest X-ray medical information market (MIMIC-CXR) database, 14 categories were obtained using natural language processing tags, among which 45,808 frontal chest radiographs were labeled as “atelectasis,” and 75,455 chest radiographs labeled “no finding.” A total of 60,000 images were extracted, including positive images labeled “atelectasis” and positive X-ray images labeled “no finding.” The data were categorized into “normal” and “atelectasis,” which were evenly distributed and randomly divided into three cohorts (training, validation, and testing) at a ratio of about 8:1:1. This retrospective study extracted 300 X-ray images labeled “atelectasis” and “normal” from patients in ICUs of The First Affiliated Hospital of Jinan University, which was labeled as an external dataset for verification in this experiment. Data set performance was evaluated using the area under the receiver operating characteristic curve (AUC), sensitivity, specificity, and positive predictive values derived from transfer learning training.

Results

It took 105 min and 6 s to train the internal training set. The AUC, sensitivity, specificity, and accuracy were 88.57, 75.10, 88.30, and 81.70%. Compared with the external validation set, the obtained AUC, sensitivity, specificity, and accuracy were 98.39, 70.70, 100, and 86.90%.

Conclusion

This study found that when detecting atelectasis, the model obtained by transfer training with sufficiently large data sets has excellent external verification and acculturate localization of lesions.

Pooled Analysis of Central Venous Pressure and Brain Natriuretic Peptide Levels in Patients With Extubation Failure

Purpose: Cardiac insufficiency has been considered to be a common cause of extubation failure. Some studies have shown that central venous pressure (CVP) and brain natriuretic peptide (BNP) are able to predict extubation outcomes. Therefore, we conducted a pooled analysis to evaluate the potential of CVP and BNP levels as predictors of extubation outcomes, using a cohort of critically ill patients who were on mechanical ventilation (MV).

Methods: We searched three online electronic databases up to October 2021. All data were analyzed using Review Manager 5.4. For each study, the analysis was performed using standardized mean differences (SMD) with 95% confidence intervals (CI).

Results: The pooled analysis of seven studies on CVP levels and extubation outcomes showed that elevated CVP levels were significantly associated with extubation failure (SMD:0.47, 95% CI: 0. 43–0.51, p < 0.00001). This association also appeared before extubation (SMD:0.47, 95% CI: 0. 43–0.51, p < 0.00001), but it did not appear after extubation (SMD: 0.63, 95% CI: −0.05–1.31, p=0.07). Similarly, pooled analysis of eight studies on BNP levels and extubation outcomes showed that increased BNP levels are closely related to extubation failure (SMD:0.68, 95% CI: 0.49–0.86, p < 0.00001). This relationship also occurs before (SMD: 0.57, 95% CI: 0.35–0.79, p < 0.00001) and after (SMD: 0.91, 95% CI: 0.59–1.23, p < 0.00001) extubation.

Conclusions: This study showed that elevated CVP and BNP levels are associated with extubation failure in critically ill patients. However, BNP levels are more valuable than CVP levels in predicting extubation outcomes.

Development and Validation of an Insulin Resistance Model for a Population with Chronic Kidney Disease Using a Machine Learning Approach

Background: Chronic kidney disease (CKD) is a complex syndrome without a definitive treatment. For these patients, insulin resistance (IR) is associated with worse renal and patient outcomes. Until now, no predictive model using machine learning (ML) has been reported on IR in CKD patients. Methods: The CKD population studied was based on results from the National Health and Nutrition Examination Survey (NHANES) of the USA from 1999 to 2012. The homeostasis model assessment of IR (HOMA-IR) was used to assess insulin resistance. We began the model building process via the ML algorithm (random forest (RF), eXtreme Gradient Boosting (XGboost), logistic regression algorithms, and deep neural learning (DNN)). We compared different receiver operating characteristic (ROC) curves from different algorithms. Finally, we used SHAP values (SHapley Additive exPlanations) to explain how the different ML models worked. Results: In this study population, 71,916 participants were enrolled. Finally, we analyzed 1,229 of these participants. Their data were segregated into the IR group (HOMA IR > 3, n = 572) or non-IR group (HOMR IR ≤ 3, n = 657). In the validation group, RF had a higher accuracy (0.77), specificity (0.81), PPV (0.77), and NPV (0.77). In the test group, XGboost had a higher AUC of ROC (0.78). In addition, XGBoost also had a higher accuracy (0.7) and NPV (0.71). RF had a higher accuracy (0.7), specificity (0.78), and PPV (0.7). In the RF algorithm, the body mass index had a much larger impact on IR (0.1654), followed by triglyceride (0.0117), the daily calorie intake (0.0602), blood HDL value (0.0587), and age (0.0446). As for the SHAP value, in the RF algorithm, almost all features were well separated to show a positive or negative association with IR. Conclusion: This was the first study using ML to predict IR in patients with CKD. Our results showed that the RF algorithm had the best AUC of ROC and the best SHAP value differentiation. This was also the first study that included both macronutrients and micronutrients. We concluded that ML algorithms, particularly RF, can help determine risk factors and predict IR in patients with CKD.

Prognostic value of glucose-to-lymphocyte ratio in critically ill patients with acute respiratory distress syndrome: A retrospective cohort study

Background

There is need to identify biomarkers for prognosis of acute respiratory distress syndrome (ADRS). This may allow early and accurate identification of patients with high‐risk ARDS to guide adjustment of clinical treatment and nursing intervention, which would ultimately improve prognosis of patients with ARDS. Biomarkers based on a combination of fasting glucose and lymphocyte counts to predict prognosis in critically ill patients with ARDS remain undefined. In this study, we investigated the association between glucose‐to‐lymphocyte ratio (GLR) and in‐hospital mortality.

Methods

The study obtained data from Medical Information Mart for Intensive Care‐IV (MIMIC‐IV Version 1.0) database. We defined the GLR as fasting glucose/lymphocyte count and the patient in‐hospital mortality was considered as the outcome. In addition, we employed linear and logistic regression models for analysis.

Results

In total, 1,085 patients with ARDS were included in this study. The eligible participants included 498 female and 587 males, with a mean age of 64.2 ± 17.5 years. Logistic regression analysis demonstrated that higher GLR was an independent risk factor for all‐cause mortality (OR =1.67, 95% CI: 1.26–2.22) after adjusting for age, sex, anion gap, white blood cell count, congestive heart failure, sequential organ failure assessment (SOFA), SBP, DBP, and respiratory rate in both the dichotomized group and subgroups. We also analyzed the in‐hospital mortality to ROC curves by comparing the value between SOFA + GLR and SOFA. The area under the curve (AUC) was 0.6991 for the SOFA + GLR (95% CI: 0.6634–0.7348), and 0.6613 for the SOFA (95% CI: 0.6238–0.6988).

Conclusion

Our data showed that the GLR was an independent predictor of in‐hospital mortality for patients with ARDS. The GLR is an integrated, readily available clinical biomarker for mortality in patients with ARDS.

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

Background

Invasive mechanical ventilation plays an important role in the prognosis of patients with sepsis. However, there are, currently, no tools specifically designed to assess weaning from invasive mechanical ventilation in patients with sepsis. The aim of our study was to develop a practical model to predict weaning in patients with sepsis.

Methods

We extracted patient information from the Medical Information Mart for Intensive Care Database-IV (MIMIC-IV) and the eICU Collaborative Research Database (eICU-CRD). Kaplan–Meier curves were plotted to compare the 28-day mortality between patients who successfully weaned and those who failed to wean. Subsequently, MIMIC-IV was divided into a training set and an internal verification set, and the eICU-CRD was designated as the external verification set. We selected the best model to simplify the internal and external validation sets based on the performance of the model.

Results

A total of 5020 and 7081 sepsis patients with invasive mechanical ventilation in MIMIC-IV and eICU-CRD were included, respectively. After matching, weaning was independently associated with 28-day mortality and length of ICU stay (p &lt; 0.001 and p = 0.002, respectively). After comparison, 35 clinical variables were extracted to build weaning models. XGBoost performed the best discrimination among the models in the internal and external validation sets (AUROC: 0.80 and 0.86, respectively). Finally, a simplified model was developed based on XGBoost, which included only four variables. The simplified model also had good predictive performance (AUROC:0.75 and 0.78 in internal and external validation sets, respectively) and was developed into a web-based tool for further review.

Conclusions

Weaning success is independently related to short-term mortality in patients with sepsis. The simplified model based on the XGBoost algorithm provides good predictive performance and great clinical applicablity for weaning, and a web-based tool was developed for better clinical application.

Predictors for extubation failure in COVID-19 patients using a machine learning approach

INTRODUCTION

Determining the optimal timing for extubation can be challenging in the intensive care. In this study, we aim to identify predictors for extubation failure in critically ill patients with COVID-19.

METHODS

We used highly granular data from 3464 adult critically ill COVID patients in the multicenter Dutch Data Warehouse, including demographics, clinical observations, medications, fluid balance, laboratory values, vital signs, and data from life support devices. All intubated patients with at least one extubation attempt were eligible for analysis. Transferred patients, patients admitted for less than 24 h, and patients still admitted at the time of data extraction were excluded. Potential predictors were selected by a team of intensive care physicians. The primary and secondary outcomes were extubation without reintubation or death within the next 7 days and within 48 h, respectively. We trained and validated multiple machine learning algorithms using fivefold nested cross-validation. Predictor importance was estimated using Shapley additive explanations, while cutoff values for the relative probability of failed extubation were estimated through partial dependence plots.

RESULTS

A total of 883 patients were included in the model derivation. The reintubation rate was 13.4% within 48 h and 18.9% at day 7, with a mortality rate of 0.6% and 1.0% respectively. The grandient-boost model performed best (area under the curve of 0.70) and was used to calculate predictor importance. Ventilatory characteristics and settings were the most important predictors. More specifically, a controlled mode duration longer than 4 days, a last fraction of inspired oxygen higher than 35%, a mean tidal volume per kg ideal body weight above 8 ml/kg in the day before extubation, and a shorter duration in assisted mode (< 2 days) compared to their median values. Additionally, a higher C-reactive protein and leukocyte count, a lower thrombocyte count, a lower Glasgow coma scale and a lower body mass index compared to their medians were associated with extubation failure.

CONCLUSION

The most important predictors for extubation failure in critically ill COVID-19 patients include ventilatory settings, inflammatory parameters, neurological status, and body mass index. These predictors should therefore be routinely captured in electronic health records.