Acute Respiratory Distress Syndrome Prediction Score: Derivation and Validation

Tree-based ensemble machine learning models in the prediction of acute respiratory distress syndrome following cardiac surgery: a multicenter cohort study

BACKGROUND

Acute respiratory distress syndrome (ARDS) after cardiac surgery is a severe respiratory complication with high mortality and morbidity. Traditional clinical approaches may lead to under recognition of this heterogeneous syndrome, potentially resulting in diagnosis delay. This study aims to develop and external validate seven machine learning (ML) models, trained on electronic health records data, for predicting ARDS after cardiac surgery.

METHODS

This multicenter, observational cohort study included patients who underwent cardiac surgery in the training and testing cohorts (data from Nanjing First Hospital), as well as those patients who had cardiac surgery in a validation cohort (data from Shanghai General Hospital). The number of important features was determined using the sliding windows sequential forward feature selection method (SWSFS). We developed a set of tree-based ML models, including Decision Tree, GBDT, AdaBoost, XGBoost, LightGBM, Random Forest, and Deep Forest. Model performance was evaluated using the area under the receiver operating characteristic curve (AUC) and Brier score. The SHapley Additive exPlanation (SHAP) techinque was employed to interpret the ML model. Furthermore, a comparison was made between the ML models and traditional scoring systems. ARDS is defined according to the Berlin definition.

RESULTS

A total of 1996 patients who had cardiac surgery were included in the study. The top five important features identified by the SWSFS were chronic obstructive pulmonary disease, preoperative albumin, central venous pressure_T4, cardiopulmonary bypass time, and left ventricular ejection fraction. Among the seven ML models, Deep Forest demonstrated the best performance, with an AUC of 0.882 and a Brier score of 0.809 in the validation cohort. Notably, the SHAP values effectively illustrated the contribution of the 13 features attributed to the model output and the individual feature's effect on model prediction. In addition, the ensemble ML models demonstrated better performance than the other six traditional scoring systems.

CONCLUSIONS

Our study identified 13 important features and provided multiple ML models to enhance the risk stratification for ARDS after cardiac surgery. Using these predictors and ML models might provide a basis for early diagnostic and preventive strategies in the perioperative management of ARDS patients.

Incidence, mortality, and predictive factors associated with acute respiratory distress syndrome in multiple trauma patients living in high-altitude areas: a retrospective study in Shigatse

Background

Acute respiratory distress syndrome (ARDS) is a severe complication that can lead to fatalities in multiple trauma patients. Nevertheless, the incidence rate and early prediction of ARDS among multiple trauma patients residing in high-altitude areas remain unknown.

Methods

This study included a total of 168 multiple trauma patients who received treatment at Shigatse People's Hospital Intensive Care Unit (ICU) between January 1, 2019 and December 31, 2021. The clinical characteristics of the patients and the incidence rate of ARDS were assessed. Univariable and multivariable logistic regression models were employed to identify potential risk factors for ARDS, and the predictive effects of these risk factors were analyzed.

Results

In the high-altitude area, the incidence of ARDS among multiple trauma patients was 37.5% (63/168), with a hospital mortality rate of 16.1% (27/168). Injury Severity Score (ISS) and thoracic injuries were identified as significant predictors for ARDS using the logistic regression model, with an area under the curve (AUC) of 0.75 and 0.75, respectively. Furthermore, a novel predictive risk score combining ISS and thoracic injuries demonstrated improved predictive ability, achieving an AUC of 0.82.

Conclusions

This study presents the incidence of ARDS in multiple trauma patients residing in the Tibetan region, and identifies two critical predictive factors along with a risk score for early prediction of ARDS. These findings have the potential to enhance clinicians' ability to accurately assess the risk of ARDS and proactively prevent its onset.

Association between cardiopulmonary bypass time and mortality among patients with acute respiratory distress syndrome after cardiac surgery

BACKGROUND

Cardiopulmonary bypass (CPB) can lead to lung injury and even acute respiratory distress syndrome (ARDS) through triggering systemic inflammatory response. The objective of this study was to investigate the impact of CPB time on clinical outcomes in patients with ARDS after cardiac surgery.

METHODS

Totally, patients with ARDS after cardiac surgery in Beijing Anzhen Hospital from January 2005 to December 2015 were retrospectively included and were further divided into three groups according to the median time of CPB. The primary endpoints were the ICU mortality and in-hospital mortality, and ICU and hospital stay. Restricted cubic spline (RCS), logistic regression, cox regression model, and receiver operating characteristic (ROC) curve were adopted to explore the relationship between CPB time and clinical endpoints.

RESULTS

A total of 54,217 patients underwent cardiac surgery during the above period, of whom 210 patients developed ARDS after surgery and were finally included. The ICU mortality and in-hospital mortality were 21.0% and 41.9% in all ARDS patients after cardiac surgery respectively. Patients with long CPB time (CPB time ≥ 173 min) had longer length of ICU stay (P = 0.011), higher ICU (P < 0.001) mortality and in-hospital(P = 0.002) mortality compared with non-CPB patients (CPB = 0). For each ten minutes increment in CPB time, the hazards of a worse outcome increased by 13.3% for ICU mortality and 9.3% for in-hospital mortality after adjusting for potential factors. ROC curves showed CPB time presented more satisfactory power to predict mortality compared with APCHEII score. The optimal cut-off value of CPB time were 160.5 min for ICU mortality and in-hospital mortality.

CONCLUSIONS

Our findings demonstrated the significant prognostic value of CPB time in patients with ARDS after cardiac surgery. Longer time of CPB was associated with poorer clinical outcomes, and could be served as an indicator to predict short-term mortality in patients with ARDS after cardiac surgery.

A deep learning model for predicting COVID-19 ARDS in critically ill patients

Background

The coronavirus disease 2019 (COVID-19) is an acute infectious pneumonia caused by a severe acute respiratory syndrome coronavirus 2 (SARS-CoV-2) infection previously unknown to humans. However, predictive studies of acute respiratory distress syndrome (ARDS) in patients with COVID-19 are limited. In this study, we attempted to establish predictive models to predict ARDS caused by COVID-19 via a thorough analysis of patients' clinical data and CT images.

Method

The data of included patients were retrospectively collected from the intensive care unit in our hospital from April 2022 to June 2022. The primary outcome was the development of ARDS after ICU admission. We first established two individual predictive models based on extreme gradient boosting (XGBoost) and convolutional neural network (CNN), respectively; then, an integrated model was developed by combining the two individual models. The performance of all the predictive models was evaluated using the area under receiver operating characteristic curve (AUC), confusion matrix, and calibration plot.

Results

A total of 103 critically ill COVID-19 patients were included in this research, of which 23 patients (22.3%) developed ARDS after admission; five predictive variables were selected and further used to establish the machine learning models, and the XGBoost model yielded the most accurate predictions with the highest AUC (0.94, 95% CI: 0.91-0.96). The AUC of the CT-based convolutional neural network predictive model and the integrated model was 0.96 (95% CI: 0.93-0.98) and 0.97 (95% CI: 0.95-0.99), respectively.

Conclusion

An integrated deep learning model could be used to predict COVID-19 ARDS in critically ill patients.

Novel models for early prediction and prevention of acute respiratory distress syndrome in patients following hepatectomy: A clinical translational study based on 1,032 patients

Background

Acute respiratory distress syndrome (ARDS) is a serious organ failure and postoperative complication. However, the incidence rate, early prediction and prevention of postoperative ARDS in patients undergoing hepatectomy remain unidentified.

Methods

A total of 1,032 patients undergoing hepatectomy between 2019 and 2020, at the Eastern Hepatobiliary Surgery Hospital were included. Patients in 2019 and 2020 were used as the development and validation cohorts, respectively. The incidence rate of ARDS was assessed. A logistic regression model and a least absolute shrinkage and selection operator (LASSO) regression model were used for constructing ARDS prediction models.

Results

The incidence of ARDS was 8.8% (43/490) in the development cohort and 5.7% (31/542) in the validation cohort. Operation time, postoperative aspartate aminotransferase (AST), and postoperative hemoglobin (Hb) were all critical predictors identified by the logistic regression model, with an area under the curve (AUC) of 0.804 in the development cohort and 0.752 in the validation cohort. Additionally, nine predictors were identified by the LASSO regression model, with an AUC of 0.848 in the development cohort and 0.786 in the validation cohort.

Conclusion

We reported the incidence of ARDS in patients undergoing hepatectomy and developed two simple and practical prediction models for early predicting postoperative ARDS in patients undergoing hepatectomy. These tools may improve clinicians' ability to early estimate the risk of postoperative ARDS and timely prevent its emergence.

Dynamic Monitoring of Serum Protein in Acute Respiratory Distress Syndrome Based on Artificial Neural Network

Acute respiratory distress syndrome (ARDS) is one of the more serious diseases in human lung disease. Reducing its incidence rate is an important task in current clinical research. Dynamic monitoring of serum protein in patients will help to achieve the early diagnosis and treatment of ARDS. In this study, a protein monitoring model based on artificial neural network is proposed. First, surface enhanced laser desorption ionization time-of-flight mass spectrometry is used for protein detection, and then BP neural network is used for protein classification and content analysis. In the experimental analysis, serum samples from patients with acute respiratory distress syndrome in our hospital from November 2020 to August 2021 were selected for experimental testing. The experimental results show that the serum protein monitoring model based on BP neural network has low error and high convergence ability and can monitor individual protein in protein monitoring, and the area under the ROC curve in diagnostic performance reaches 0.854. The above results show that the artificial neural network has a good effect on the dynamic monitoring of serum protein in acute respiratory distress syndrome, and the diagnostic performance evaluation can reach 0.854, which has the ability to significantly improve the clinical diagnosis and treatment of acute respiratory distress syndrome.