RESUMEN
Background Patients with chronic critical illness (CCI) experience poor prognoses and incur high medical costs. However, there is currently limited clinical awareness of sepsis-associated CCI, resulting in insufficient vigilance. Therefore, it is necessary to build a machine learning model that can predict whether sepsis patients will develop CCI. Methods Clinical data on 19,077 sepsis patients from the Medical Information Mart for Intensive Care IV (MIMIC-IV) database were analyzed. Predictive factors were identified using the Student's t-test, Mann-Whitney U test, or χ 2 test. Six machine learning classification models, namely, the logistic regression, support vector machine, decision tree, random forest, extreme gradient enhancement, and artificial neural network, were established. The optimal model was selected on the basis of its performance. Calibration curves were used to evaluate the accuracy of model classification, while the external validation dataset was used to evaluate the performance of the model. Results Thirty-seven characteristics, such as elevated alanine aminotransferase, rapid heart rate, and high Logistic Organ Dysfunction System scores, were identified as risk factors for developing CCI. The area under the receiver operating characteristic curve (AUROC) values for all models were above 0.73 on the internal test set. Among them, the extreme gradient enhancement model exhibited superior performance (F1 score = 0.91, AUROC = 0.91, Brier score = 0.052). It also exhibited stable prediction performance on the external validation set (AUROC = 0.72). Conclusion A machine learning model was established to predict whether sepsis patients will develop CCI. It can provide useful predictive information for clinical decision-making.
RESUMEN
Background: Cuproptosis is a copper-dependent cell death that is connected to the development and immune response of multiple diseases. However, the function of cuproptosis in the immune characteristics of sepsis remains unclear. Method: We obtained two sepsis datasets (GSE9960 and GSE134347) from the GEO database and classified the raw data with R packages. Cuproptosis-related genes were manually curated, and differentially expressed cuproptosis-related genes (DECuGs) were identified. Afterwards, we applied enrichment analysis and identified key DECuGs by performing machine learning techniques. Then, the immune cell infiltrations and correlation between DECuGs and immunocyte features were created by the CIBERSORT algorithm. Subsequently, unsupervised hierarchical clustering analysis was performed based on key DECuGs. We then constructed a ceRNA network based on key DECuGs by using multi-step computational strategies and predicted potential drugs in the DrugBank database. Finally, the role of these key genes in immune cells was validated at the single-cell RNA level between septic patients and healthy controls. Results: Overall, 16 DECuGs were obtained, and most of them had lower expression levels in sepsis samples. Afterwards, we obtained six key DECuGs by performing machine learning. Then, the LIPT1-T-cell CD4 memory resting was the most positively correlated DECuG-immunocyte pair. Subsequently, two different subclusters were identified by six DECuGs. Bioinformatics analysis revealed that there were different immune characteristics between the two subclusters. Moreover, we identified the key lncRNA OIP5-AS1 within the ceRNA network and obtained 4 drugs that may represent novel drugs for sepsis. Finally, these key DECuGs were statistically significantly dysregulated in another validation set and showed a major distribution in monocytes, T cells, B cells, NK cells and platelets at the single-cell RNA level. Conclusion: These findings suggest that cuproptosis might promote the progression of sepsis by affecting the immune system and metabolic dysfunction, which provides a new direction for understanding potential pathogenic processes and therapeutic targets in sepsis.