Prospective model for predicting renal recovery in cardiac surgery patients with acute kidney injury requiring renal replacement therapy

Development and validation of a nomogram model for central venous access device-related thrombosis in hospitalized children

This study aimed to develop and validate a nomogram model of central venous access device-related thrombosis (CRT) for hospitalized children. A total of 503 consecutive cases from a hospital in Changsha City, Hunan Province were stochastically classified into the training set and internal validation set at a ratio of 7:3, and 85 consecutive cases in two hospitals in Urumqi City, Xinjiang Uygur Autonomous Region were collected as an external validation set. Univariate analysis and multivariate analysis on CRT-related risk factors of hospitalized children were conducted, a logistic regression model was employed to establish the nomogram, and the discrimination, calibration, and decision curve analysis was performed to assess the proposed nomogram model. The nomogram model involved seven independent risk factors, including blind catheterization, abnormal liver function, central line-associated bloodstream infection, infection, number of catheter lines, leukemia, and bed rest > 72 h. The discrimination results showed that the area under the receiver operating characteristic curve of the training set, internal validation set, and external validation set was 0.74, 0.71, and 0.76 respectively, and the accuracy rates of the proposed nomogram model were 79%, 72%, and 71% in the training set, internal validation set, and external validation set. The calibration results also showed that the calibration curve had great fitness for each dataset. More importantly, the decision curve suggested that the proposed nomogram model had a prominent clinical significance.

CONCLUSION

The nomogram model can be used as a risk assessment tool to reduce the missed diagnosis rate and the incidence of CRT in hospitalized children.

WHAT IS KNOWN

• Central venous access device-related thrombosis is generally asymptomatic for hospitalized children, causing the missed diagnosis of central venous access device-related thrombosis easily. • No risk prediction nomogram model for central venous access device-related thrombosis in hospitalized children has been established.

WHAT IS NEW

• A visual and personalized nomogram model was built by seven accessible variables (blind catheterization, abnormal liver function, central line-associated bloodstream infection, infection, number of catheter lines, leukemia, and bed rest > 72 h). • The model can effectively predict the risk of central venous access device-related thrombosis for hospitalized children.

Year in Review 2021: Noteworthy Literature in Cardiothoracic Critical Care

This year marked a number of milestones in critical care. As vaccines for the SARS-CoV-2 virus became widely available and were confirmed to be exceptionally effective against severe illness and hospitalization, we were then faced with new variants and the resource-intense responses necessary to combat them. Despite challenges new and old, we have persevered and continued to provide excellent care to our patients while pushing the boundaries of clinical research. This article is a collection of studies published in 2021 relevant to critical care, with a specific focus on cardiothoracic critical care. To ignore the impact of the COVID-19 pandemic would do a disservice to our colleagues, many of whom have made incredible breakthroughs in novel therapies to the coronavirus, and yet we present additional themes of delirium, acute kidney injury, lung transplant, advances in ECMO as well as biomarkers of sepsis.

Application of Machine Learning to Predict Acute Kidney Disease in Patients With Sepsis Associated Acute Kidney Injury

Background: Sepsis-associated acute kidney injury (AKI) is frequent in patients admitted to intensive care units (ICU) and may contribute to adverse short-term and long-term outcomes. Acute kidney disease (AKD) reflects the adverse events developing after AKI. We aimed to develop and validate machine learning models to predict the occurrence of AKD in patients with sepsis-associated AKI. Methods: Using clinical data from patients with sepsis in the ICU at Beijing Friendship Hospital (BFH), we studied whether the following three machine learning models could predict the occurrence of AKD using demographic, laboratory, and other related variables: Recurrent Neural Network-Long Short-Term Memory (RNN-LSTM), decision trees, and logistic regression. In addition, we externally validated the results in the Medical Information Mart for Intensive Care III (MIMIC III) database. The outcome was the diagnosis of AKD when defined as AKI prolonged for 7-90 days according to Acute Disease Quality Initiative-16. Results: In this study, 209 patients from BFH were included, with 55.5% of them diagnosed as having AKD. Furthermore, 509 patients were included from the MIMIC III database, of which 46.4% were diagnosed as having AKD. Applying machine learning could successfully achieve very high accuracy (RNN-LSTM AUROC = 1; decision trees AUROC = 0.954; logistic regression AUROC = 0.728), with RNN-LSTM showing the best results. Further analyses revealed that the change of non-renal Sequential Organ Failure Assessment (SOFA) score between the 1st day and 3rd day (Δnon-renal SOFA) is instrumental in predicting the occurrence of AKD. Conclusion: Our results showed that machine learning, particularly RNN-LSTM, can accurately predict AKD occurrence. In addition, Δ SOFAnon-renal plays an important role in predicting the occurrence of AKD.