Prediction of acute kidney injury after cardiac surgery from preoperative N-terminal pro-B-type natriuretic peptide

AKIMLpred: An interpretable machine learning model for predicting acute kidney injury within seven days in critically ill patients based on a prospective cohort study

BACKGROUND

Early recognition and timely intervention for AKI in critically ill patients were crucial to reduce morbidity and mortality. This study aimed to use biomarkers to construct a optimal machine learning model for early prediction of AKI in critically ill patients within seven days.

METHODS

The prospective cohort study enrolled 929 patients altogether who were admitted in ICU including 680 patients in training set (Jiefang Campus) and 249 patients in external testing set (Binjiang Campus). After performing strict inclusion and exclusion criteria, 421 patients were selected in training set for constructing predictive model and 167 patients were selected in external testing for evaluating the predictive performance of resulting model. Urine and blood samples were collected for kidney injury associated biomarkers detection. Baseline clinical information and laboratory data of the study participants were collected. We determined the average prediction efficiency of six machine learning models through 10-fold cross validation.

RESULTS

In total, 78 variables were collected when admission in ICU and 43 variables were statistically significant between AKI and non-AKI cohort. Then, 35 variables were selected as independent features for AKI by univariate logistic regression. Spearman correlation analysis was used to remove two highly correlated variables. Three ranking methods were used to explore the influence of 33 variables for further determining the best combination of variables. The gini importance ranking method was found to be applicable for variables filtering. The predictive performance of AKIMLpred which constructed by the XGBoost algorithm was the best among six machine learning models. When the AKIMLpred included the nine features (NGAL, IGFBP7, sCysC, CAF22, KIM-1, NT-proBNP, IL-6, IL-18 and L-FABP) with the highest influence ranking, its model had the best prediction performance, with an AUC of 0.881 and an accuracy of 0.815 in training set, similarly, with an AUC of 0.889 and an accuracy of 0.846 in validation set. Moreover, the performace was slightly outperformed in testing set with an AUC of 0.902 and an accuracy of 0.846. The SHAP algorithm was used to interpret the prediction results of AKIMLpred. The web-calculator of AKIMLpred was shown for predicting AKI with more convenient(https://www.xsmartanalysis.com/model/list/predict/model/html?mid=8065&symbol=11gk693982SU6AE1ms21). AKIMLpred was better than the optimal model built with only routine tests for predicting AKI in critically ill patients within 7 days.

CONCLUSION

The model AKIMLpred constructed by the XGBoost algorithm with selecting the nine most influential biomarkers in the gini importance ranking method had the best performance in predicting AKI in critically ill patients within 7 days. This data-driven predictive model will help clinicians to make quick and accurate diagnosis.

Risk Prediction Models for Renal Function Decline After Cardiac Surgery Within Different Preoperative Glomerular Filtration Rate Strata

BACKGROUND

Our goal was to create a simple risk-prediction model for renal function decline after cardiac surgery to help focus renal follow-up efforts on patients most likely to benefit.

METHODS AND RESULTS

This single-center retrospective cohort study enrolled 24 904 patients who underwent cardiac surgery from 2012 to 2019 at Fuwai Hospital, Beijing, China. An estimated glomerular filtration rate (eGFR) reduction of ≥30% 3 months after surgery was considered evidence of renal function decline. Relative to patients with eGFR 60 to 89 mL/min per 1.73 m2 (4.5% [531/11733]), those with eGFR ≥90 mL/min per 1.73 m2 (10.9% [1200/11042]) had a higher risk of renal function decline, whereas those with eGFR ≤59 mL/min per 1.73 m2 (5.8% [124/2129]) did not. Each eGFR stratum had a different strongest contributor to renal function decline: increased baseline eGFR levels for patients with eGFR ≥90 mL/min per 1.73 m2, transfusion of any blood type for patients with eGFR 60 to 89 mL/min per 1.73 m2, and no recovery of renal function at discharge for patients with eGFR ≤59 mL/min per 1.73 m2. Different nomograms were established for the different eGFR strata, which yielded a corrected C-index value of 0.752 for eGFR ≥90 mL/min per 1.73 m2, 0.725 for eGFR 60-89 mL/min per 1.73 m2 and 0.791 for eGFR ≤59 mL/min per 1.73 m2.

CONCLUSIONS

Predictors of renal function decline over the follow-up showed marked differences across the eGFR strata. The nomograms incorporated a small number of variables that are readily available in the routine cardiac surgical setting and can be used to predict renal function decline in patients stratified by baseline eGFR.

Usefulness of the N-Terminal of the Prohormone Brain Natriuretic Peptide in Predicting Acute Kidney Injury Requiring Renal Replacement Therapy in Patients Undergoing Heart Valve Surgery

Background and Objectives: By definition, acute kidney injury (AKI) is a clinical syndrome diagnosed when the increase in serum creatinine concentration is >0.3 mg/dL in 48 h or >1.5-fold in the last seven days or when diuresis < 0.5 mL/kg/h for a consecutive 6 h. AKI is one of the severe complications that may occur in the early postoperative period in patients undergoing heart valve surgery, significantly increasing the risk of death. Early implementation of renal replacement therapy increases the chances of improving treatment results in patients with postoperative AKI. The study assessed the predictive ability of selected preoperative and perioperative parameters for the occurrence of postoperative AKI requiring renal replacement therapy in the early postoperative period in a group of patients with severe valvular heart disease. Materials and Methods: A prospective study was conducted on a group of patients undergoing consecutive heart valve surgeries. The primary endpoint was postoperative AKI requiring renal replacement therapy. AKI was diagnosed with an increase in serum creatinine > 0.3 mg/dL in 48 h or >1.5-fold in the previous 7 days and/or a decrease in diuresis < 0.5 mL/kg/h for 6 h. The observation period was until the patient was discharged home or death occurred. Logistic regression analysis was used to assess which variables were predictive of primary endpoint, and odds ratios (OR) were calculated with a 95% confidence interval (CI). Multivariate analysis was based on the result of single factor logistic regression, i.e., to further steps, all statistically significant variables were taken into consideration. Results: A total of 607 patients were included in the study. The primary endpoint occurred in 50 patients. At multivariate analysis: NT-proBNP (OR 1.406; 95% CI 1.015-1.949; p = 0.04), CRP (OR 1.523; 95% CI 1.171-1.980; p = 0.001), EuroSCORE II (OR 1.090; 95% CI 1.014-1.172; p = 0.01), age (OR 1.037; 95% CI 1.001-1.075; p = 0.04) and if they stayed in the intensive care unit longer than 2 days (OR 9.077; 95% CI 2.026-40.663; p = 0.004) remained the independent predictors of the primary endpoint. The mean preoperative NT-proBNP level was 2063 pg/mL (±1751). Thirty-eight patients with AKI requiring renal replacement therapy died in intrahospital follow-up. Conclusions: The results of the presented study indicate that a high preoperative level of NT-proBNP and postoperative hemodynamic instability may be associated with a significant risk of a postoperative AKI requiring renal replacement therapy. The results of the study may also suggest that qualifying for heart valve surgery earlier may be associated with improved prognosis in this group of patients.

Prognostic Value of Natriuretic Peptide Levels for Adverse Renal Outcomes in Patients With Moderate to Severe Acute Kidney Injury With or Without Heart Failure

Background Natriuretic peptides have been recommended as biomarkers for the diagnosis and prognosis of patients with heart failure and are often elevated in the setting of acute kidney injury. We sought to demonstrate the associations between increased baseline NT-proBNP (N-terminal pro-B-type natriuretic peptide) and adverse renal outcomes in patients with moderate-to-severe acute kidney injury. Methods and Results We reviewed electronic medical records of consecutive patients with acute kidney injury stage 2 and 3 admitted to the Cleveland Clinic between September 2011 and December 2021. Patients with NT-proBNP levels collected before renal consultation or dialysis initiation were included. Adverse renal outcomes included dialysis requirement and dialysis dependence defined as patients undergoing dialysis within 72 hours before hospital discharge or in-hospital mortality. In our study cohort (n=3811), 2521 (66%) patients underwent dialysis, 1619 (42%) patients became dialysis dependent, and 1325 (35%) patients had in-hospital mortality. After adjusting for cardiorenal risk factors, compared with the lowest quartile, the highest quartile of NT-proBNP (≥18 215 pg/mL) was associated with increased likelihood of dialysis requirement (adjusted odds ratio [OR], 2.36 [95% CI, 1.87-2.99]), dialysis dependence (adjusted OR, 1.89 [95% CI, 2.53-1.34]), and in-hospital mortality (adjusted OR, 1.34 [95% CI, 1.01-1.34]). Conclusions Increased NT-proBNP was associated with an increased risk of dialysis requirement, becoming dialysis dependent, and in-hospital mortality in patients with moderate-to-severe acute kidney injury.

Interpretable machine learning models for early prediction of acute kidney injury after cardiac surgery

OBJECTIVE

Postoperative acute kidney injury (PO-AKI) is a common complication after cardiac surgery. We aimed to evaluate whether machine learning algorithms could significantly improve the risk prediction of PO-AKI.

METHODS

The retrospective cohort study included 2310 adult patients undergoing cardiac surgery in a tertiary teaching hospital, China. Postoperative AKI and severe AKI were identified by the modified KDIGO definition. The sample was randomly divided into a derivation set and a validation set based on a ratio of 4:1. Exploiting conventional logistic regression (LR) and five ML algorithms including decision tree, random forest, gradient boosting classifier (GBC), Gaussian Naive Bayes and multilayer perceptron, we developed and validated the prediction models of PO-AKI. We implemented the interpretation of models using SHapley Additive exPlanation (SHAP) analysis.

RESULTS

Postoperative AKI and severe AKI occurred in 1020 (44.2%) and 286 (12.4%) patients, respectively. Compared with the five ML models, LR model for PO-AKI exhibited the largest AUC (0.812, 95%CI: 0.756, 0.860, all P < 0.05), sensitivity (0.774, 95%CI: 0.719, 0.813), accuracy (0.753, 95%CI: 0.719, 0.781) and Youden index (0.513, 95%CI: 0.451, 0.573). Regarding severe AKI, GBC algorithm showed a significantly higher AUC than the other four ML models (all P < 0.05). Although no significant difference (P = 0.173) was observed in AUCs between GBC (0.86, 95%CI: 0.808, 0.902) and conventional logistic regression (0.803, 95%CI: 0.746, 0.852), GBC achieved greater sensitivity, accuracy and Youden index than conventional LR. Notably, SHAP analyses showed that preoperative serum creatinine, hyperlipidemia, lipid-lowering agents and assisted ventilation time were consistently among the top five important predictors for both postoperative AKI and severe AKI.

CONCLUSION

Logistic regression and GBC algorithm demonstrated moderate to good discrimination and superior performance in predicting PO-AKI and severe AKI, respectively. Interpretation of the models identified the key contributors to the predictions, which could potentially inform clinical interventions.

Effect of preoperative risk on the association between intraoperative hypotension and postoperative acute kidney injury in cardiac surgery

BACKGROUND

Acute kidney injury (AKI), a common and severe complication after cardiac surgery, has been demonstrated to be associated with intraoperative hypotension (IOH). The reproducibility of this finding and whether preoperative risk modifies the association remain unclear. We hypothesised that the relationship between IOH and AKI after cardiac surgery varies by preoperative risk.

METHODS

We conducted a single-centre, retrospective cohort study to analyse the association between IOH and postoperative AKI by stratifying patients using preoperative risk factors. IOH was defined as a mean arterial pressure (MAP) of less than 65 mmHg and characterised by the cumulative duration and area under the curve (AUC).

RESULTS

Ten variables could be identified as risk factors: age, smoking status, NYHA III/Ⅳ, emergency surgery, peripheral vascular disease, cerebrovascular disease, heart failure, hypertension, previous cardiac surgery, and NT-proBNP concentration. The risk prediction model divided the patients into three equal-sized preoperative risk groups. Low-risk patients demonstrated no association between AKI and IOH of any severity, while high-risk patients demonstrated a statistically significant association between AKI and IOH with a cumulative duration greater than 104 min (adjusted odds ratio [OR]: 2.27, 95% confidence interval [CI]: 1.10-4.74; and adjusted OR: 3.63, 95% CI: 1.77-7.58) and an AUC greater than 905 mmHg min (adjusted OR: 2.08, 95% CI: 1.01-4.36; and adjusted OR: 4.00, 95% CI: 1.95-8.43).

CONCLUSION

IOH is a significant independent risk factor for AKI after cardiac surgery. Patients with higher baseline risk showed a more prominent relationship between IOH and postoperative AKI than low-risk patients.

Incidence and predictors of postoperative acute kidney injury in older adults with hip fractures

OBJECTIVES

Main Purpose: To clarify the incidence and predictors of acute kidney injury (AKI) after hip fracture surgery; Secondary Purpose: To investigate the impact of AKI on the length of stay (LOS) and mortality of patients.

METHODS

We retrospectively evaluated data from 644 hip fracture patients at Peking University First Hospital from 2015 to 2021, and divided the patients into AKI and Non-AKI groups according to whether AKI occurred after surgery. Logistic regression was used to clarify the risk factors for AKI, draw ROC curves, and analyze the odds ratio (OR) for LOS and death at 30 days, 3 months, and 1 year for patients with AKI.

RESULTS

The prevalence of AKI after hip fracture was 12.1%. Age, BMI, and postoperative brain natriuretic peptide (BNP) levels were risk factors for AKI after hip fracture surgery. The risk of AKI in underweight patients, overweight patients and obese patients was 2.24, 1.89, and 2.58 times. Compared to patients with BNP levels <800 pg/ml, the risk of AKI was 22.34-fold for postoperative BNP levels>1500 pg/ml. The risk of a one-grade increase in LOS was 2.84 times higher in the AKI group and the mortality of patients with AKI were higher.

CONCLUSION

The incidence of AKI after hip fracture surgery was 12.1%. Advanced age, low BMI, and postoperative high level BNP were risk factors for AKI. Surgeons need to pay more attention to patients with older age, low BMI and high postoperative BNP levels in order to proactively prevent the development of postoperative AKI.

A novel model for predicting a composite outcome of major complications after valve surgery

Background

On-pump valve surgeries are associated with high morbidity and mortality. The present study aimed to reliably predict a composite outcome of postoperative complications using a minimum of easily accessible clinical parameters.

Methods

A total of 7,441 patients who underwent valve surgery were retrospectively analyzed. Data for 6,220 patients at West China Hospital of Sichuan University were used to develop a predictive model, which was validated using data from 1,221 patients at the Second Affiliated Hospital of Zhejiang University School of Medicine. The primary outcome was a composite of major complications: all-cause death in hospital, stroke, myocardial infarction, and severe acute kidney injury. The predictive model was constructed using the least absolute shrinkage and selection operator as well as multivariable logistic regression. The model was assessed in terms of the areas under receiver operating characteristic curves, calibration, and decision curve analysis.

Results

The primary outcome occurred in 129 patients (2.1%) in the development cohort and 71 (5.8%) in the validation cohort. Six variables were retained in the predictive model: New York Heart Association class, diabetes, glucose, blood urea nitrogen, operation time, and red blood cell transfusion during surgery. The C-statistics were 0.735 (95% CI, 0.686-0.784) in the development cohort and 0.761 (95% CI, 0.694-0.828) in the validation cohort. For both cohorts, calibration plots showed good agreement between predicted and actual observations, and ecision curve analysis showed clinical usefulness. In contrast, the well-established SinoSCORE did not accurately predict the primary outcome in either cohort.

Conclusions

This predictive nomogram based on six easily accessible variables may serve as an "early warning" system to identify patients at high risk of major complications after valve surgery.

Clinical Trial Registration

[www.ClinicalTrials.gov], identifier [NCT04476134].

Development and Validation of a Machine Learning Predictive Model for Cardiac Surgery-Associated Acute Kidney Injury

OBJECTIVE

We aimed to develop and validate a predictive machine learning (ML) model for cardiac surgery associated with acute kidney injury (CSA-AKI) based on a multicenter randomized control trial (RCT) and a Medical Information Mart for Intensive Care-IV (MIMIC-IV) dataset.

METHODS

This was a subanalysis from a completed RCT approved by the Ethics Committee of Fuwai Hospital in Beijing, China (NCT03782350). Data from Fuwai Hospital were randomly assigned, with 80% for the training dataset and 20% for the testing dataset. The data from three other centers were used for the external validation dataset. Furthermore, the MIMIC-IV dataset was also utilized to validate the performance of the predictive model. The area under the receiver operating characteristic curve (ROC-AUC), the precision-recall curve (PR-AUC), and the calibration brier score were applied to evaluate the performance of the traditional logistic regression (LR) and eleven ML algorithms. Additionally, the Shapley Additive Explanations (SHAP) interpreter was used to explain the potential risk factors for CSA-AKI.

RESULT

A total of 6495 eligible patients undergoing cardiopulmonary bypass (CPB) were eventually included in this study, 2416 of whom were from Fuwai Hospital (Beijing), for model development, 562 from three other cardiac centers in China, and 3517 from the MIMICIV dataset, were used, respectively, for external validation. The CatBoostClassifier algorithms outperformed other models, with excellent discrimination and calibration performance for the development, as well as the MIMIC-IV, datasets. In addition, the CatBoostClassifier achieved ROC-AUCs of 0.85, 0.67, and 0.77 and brier scores of 0.14, 0.19, and 0.16 in the testing, external, and MIMIC-IV datasets, respectively. Moreover, the utmost important risk factor, the N-terminal brain sodium peptide (NT-proBNP), was confirmed by the LASSO method in the feature section process. Notably, the SHAP explainer identified that the preoperative blood urea nitrogen level, prothrombin time, serum creatinine level, total bilirubin level, and age were positively correlated with CSA-AKI; preoperative platelets level, systolic and diastolic blood pressure, albumin level, and body weight were negatively associated with CSA-AKI.

CONCLUSIONS

The CatBoostClassifier algorithms outperformed other ML models in the discrimination and calibration of CSA-AKI prediction cardiac surgery with CPB, based on a multicenter RCT and MIMIC-IV dataset. Moreover, the preoperative NT-proBNP level was confirmed to be strongly related to CSA-AKI.

Early Postoperative Acetaminophen Administration and Severe Acute Kidney Injury After Cardiac Surgery

RATIONALE & OBJECTIVE

Oxidative stress may contribute to the development of acute kidney injury (AKI) after cardiac surgery. Acetaminophen can be considered an antioxidant because it inhibits hemoprotein-catalyzed lipid peroxidation. We hypothesized that perioperative acetaminophen administration is associated with reduced AKI after cardiac surgery.

STUDY DESIGN

Retrospective observational cohort study.

SETTING & PARTICIPANTS

Patients aged≥18 years who had cardiac surgery were identified from 2 publicly available clinical registries: the Medical Information Mart for Intensive Care III (MIMIC-III) and the eICU Collaborative Research Database (eICU).

EXPOSURE

Administration of acetaminophen in the first 48 hours after surgery.

OUTCOME

Severe AKI in the first 7 days after surgery, defined as stage 2 or stage 3 AKI according to KDIGO criteria.

ANALYTICAL APPROACH

Multivariable cause-specific hazards regression analysis.

RESULTS

We identified 5,791 patients from the MIMIC-III and 3,840 patients from the eICU registries. The overall incidence of severe AKI was 58% (3,390 patients) in the MIMIC-III cohort and 37% (1,431 patients) in the eICU cohort. Acetaminophen was administered in the early postoperative period to 4,185 patients (72%) and 2,737 patients (71%) in these 2 cohorts, respectively. In multivariable regression models, early postoperative use of acetaminophen was associated with a lower risk of severe AKI in both the MIMIC-III (adjusted hazard ratio [AHR], 0.86 [95% CI, 0.79-0.94]) and eICU (AHR, 0.84 [95% CI, 0.72-0.97]) cohorts. The benefit was consistent across sensitivity and subgroup analyses.

LIMITATIONS

No data on acetaminophen dose.

CONCLUSIONS

Early postoperative acetaminophen administration was independently associated with a lower risk of severe AKI in adults recovering from cardiac surgery. Prospective trials are warranted to assess the extent to which the observed association is causal and estimate the extent to which acetaminophen administration might prevent or reduce the severity of AKI.

Using preoperative N-terminal pro-B-type natriuretic peptide levels for predicting major adverse cardiovascular events and myocardial injury after noncardiac surgery in Chinese advanced-age patients

BACKGROUND

N-terminal pro-B-type natriuretic peptide (NT-proBNP) is often viewed as an indicator for heart failure. However, the prognostic association and the predictive utility of NT-proBNP for postoperative major adverse cardiovascular events (MACEs) and myocardial injury after noncardiac surgery (MINS) among older patients are unclear.

METHODS

In this study, we included 5033 patients aged 65 years or older who underwent noncardiac surgery with preoperative NT-proBNP recorded. Logistic regression was adopted to model the associations between preoperative NT-proBNP and the risk of MACEs and MINS. The receiver operating characteristic curve was used to determine the predictive value of NT-proBNP.

RESULTS

A total of 5033 patients were enrolled, 63 patients (1.25%) and 525 patients (10.43%) had incident postoperative MACEs and MINS, respectively. Analysis of the receiver operating characteristic curve indicated that the cutoff values of ln (NT-proBNP) for MACEs and MINS were 5.16 (174 pg/mL) and 5.30 (200 pg/mL), respectively. Adding preoperative ln (NT-proBNP) to the Revised Cardiac Risk Index score and the Cardiac and Stroke Risk Model boosted the area under the receiver operating characteristic curves from 0.682 to 0.726 and 0.787 to 0.804, respectively. The inclusion of preoperative NT-proBNP in the prediction models significantly increased the reclassification and discrimination.

CONCLUSIONS

Increased preoperative NT-proBNP was associated with a higher risk of postoperative MACEs and MINS. The inclusion of NT-proBNP enhances the predictive ability of the preexisting models.

NT-proBNP as a Predictive Biomarker for Contrast-Induced Nephropathy in ACS Patients Undergoing Coronary Angiogram – An Observational Study

Objectives:

1. To assess the value of baseline NTproBNP at admission and to determine the levels of serum creatinine at 48 hours and 72 hours after procedure for evidence of contrast-induced nephropathy (CIN) for patients undergoing CAG. 2. To evaluate the relationship between the values of NTproBNP and evidence of CIN.

Materials and Methods:

This is an observational study performed between June 2021-November 2021 at Nizams Institute of Medical Sciences in 75 patients diagnosed with ACS. we assessed the role of nt pro bnp as a predictive biomarker for diagnosis of contrast induced nephropathy in patients of ACS undergoing coronary angiography. Serum creatinine is repeated at 48 h post procedure and compared to baseline.

Results:

Spearman’s correlation test was used to assess the correlation between NT-proBNP values and ejection fraction on the 2D echo. The rho value (-0.69) was suggestive of a strong negative correlation. P value & lt; 0.001 making it statistically significant. Simple linear regression analysis was used to predict the NT-proBNP levels by ejection fraction percentage among study patients, it showed that, for every 1% decrease in ejection fraction, the NT-proBNP levels will significantly increase by 102.90 pg/mL at P and lt; 0.001. Wilcoxon Signed Rank test was used to compare the baseline serum creatinine values with 48/72 h serum creatinine values after undergoing angiography with contrast, incidence of acute kidney injury (AKI) as shown by the resulting P value was and lt; 0.001, thus statistically significant. The ROC curve analysis to establish the association between NT-proBNP as a marker for incidence of AKI (CIN) shows shows that, NT-proBNP cut off and gt;1670 pg/mL has a sensitivity of 81.82% and specifity of 98.44% and is statistically significant with P value and lt; 0.001.

Conclusion:

It was observed that NT-proBNP >1670 pg/mL prior to the procedure, was significantly associated with the risk of development of contrast induced nephropathy. Measurement of serum NT-proBNP pre procedure aids in identifying at risk population for developing CIN.

Postoperative Acute Kidney Injury

Postoperative AKI is a common complication of major surgery and is associated with significant morbidity and mortality. The Kidney Disease Improving Global Outcomes AKI definition allows consensus classification and identification of postoperative AKI through changes in serum creatinine and/or urine output. However, such conventional diagnostic criteria may be inaccurate in the postoperative period, suggesting a potential to refine diagnosis by application of novel diagnostic biomarkers. Risk factors for the development of postoperative AKI can be thought of in terms of preoperative, intraoperative, and postoperative factors and, as such, represent areas that may be targeted perioperatively to minimize the risk of AKI. The treatment of postoperative AKI remains predominantly supportive, although application of management bundles may translate into improved outcomes.

Machine learning algorithms to predict major bleeding after isolated coronary artery bypass grafting

Objectives

Postoperative major bleeding is a common problem in patients undergoing cardiac surgery and is associated with poor outcomes. We evaluated the performance of machine learning (ML) methods to predict postoperative major bleeding.

Methods

A total of 1,045 patients who underwent isolated coronary artery bypass graft surgery (CABG) were enrolled. Their datasets were assigned randomly to training (70%) or a testing set (30%). The primary outcome was major bleeding defined as the universal definition of perioperative bleeding (UDPB) classes 3–4. We constructed a reference logistic regression (LR) model using known predictors. We also developed several modern ML algorithms. In the test set, we compared the area under the receiver operating characteristic curves (AUCs) of these ML algorithms with the reference LR model results, and the TRUST and WILL-BLEED risk score. Calibration analysis was undertaken using the calibration belt method.

Results

The prevalence of postoperative major bleeding was 7.1% (74/1,045). For major bleeds, the conditional inference random forest (CIRF) model showed the highest AUC [0.831 (0.732–0.930)], and the stochastic gradient boosting (SGBT) and random forest models demonstrated the next best results [0.820 (0.742–0.899) and 0.810 (0.719–0.902)]. The AUCs of all ML models were higher than [0.629 (0.517–0.641) and 0.557 (0.449–0.665)], as achieved by TRUST and WILL-BLEED, respectively.

Conclusion

ML methods successfully predicted major bleeding after cardiac surgery, with greater performance compared with previous scoring models. Modern ML models may enhance the identification of high-risk major bleeding subpopulations.

Predicting acute kidney injury after cardiac surgery: much work still to be done

Accurate preoperative risk prediction for perioperative complications such as acute kidney injury (AKI) may serve to better inform patients and families of risk before surgery, assist with resource requirement planning, and aid with cohort enrichment for enrolment into clinical trials. Where a specific risk factor is modifiable, it may offer a potential therapeutic target for risk reduction. The report by Wang and colleagues describes the modest incremental benefit of N-terminal pro brain natriuretic peptide levels when added to almost 20 other variables for the preoperative prediction of AKI after cardiac surgery. This is consistent with previous smaller studies, but there are important additional questions still to be answered before this biomarker might be used for this purpose in clinical practice.