Background This study aims to evaluate the prognostic value of a pulmonary involvement (PI) score in COVID-19 patients, both independently and in combination with clinical and laboratory parameters, following the adjustment of the dynamic zeroing policy in China. Methods A total of 288 confirmed COVID-19 pneumonia patients (mild/moderate group, 155; severe group, 133) from the Emergence Department, Beijing Chaoyang Hospital, were enrolled in this study and allocated to the training and validation cohort. The PI score of the initial chest CT was evaluated using a semi-quantitative scoring system, and clinical and laboratory parameters were collected. Radiomics and combination predictive models were developed using the least absolute shrinkage and selection operator (LASSO) logistic regression algorithm and multivariate logistic regression. The models' performance for predicting severe COVID-19 was assessed by receiver operating characteristics curve (ROC) analysis and calibration curve. Results Compared with the mild/moderate patients, the severe patients had higher levels of C-reactive protein (CRP), D-dimer, procalcitonin (PCT), and brain natriuretic peptide (BNP), but lower blood oxygen saturation and vaccination rate (P < 0.05). The severe group had a higher incidence of consolidation, multi-lobe involvement, interlobular septal thickening, air bronchogram sign, and pleural effusion compared to the mild/moderate group (P < 0.05). Moreover, the PI total score of severe patients was 16.4 ± 3.8, significantly higher than 8.5 ± 3.8 of milder patients (P < 0.001). The developed predictive nomogram, which includes four clinical characteristics and one CT feature, exhibited good performance in predicting severe COVID-19 with an area under the ROC (AUC) of 0.98 (95% CI, 0.97-1.00) in the training dataset, and 0.97 (95% CI, 0.94-1.00) in the validation dataset. Conclusions The combination predictive model, including CT score, clinical factors, and laboratory data, shows favorable predictive efficacy for severe COVID-19, which could potentially aid clinicians in triaging emergency patients.
Background Acute kidney injury (AKI) can make cases of acute respiratory distress syndrome (ARDS) more complex, and the combination of the two can significantly worsen the prognosis. Our objective, therefore, is to utilize machine learning techniques to construct models that can promptly identify the risk of AKI in ARDS patients, and provide guidance for early intervention and treatment, ultimately leading to improved prognosis. Method We obtained data regarding ARDS patients from the Medical Information Mart for Intensive Care III (MIMIC-III) database and utilized 11 machine learning (ML) algorithms to construct our predictive models. We selected the best model based on various metrics, and visualized the importance of its features using Shapley additive explanations (SHAP). We then created a more concise model using fewer variables, and optimized it using hyperparameter optimization (HPO). Additionally, we developed a web-based calculator to facilitate clinical usage. Result A total of 928 ARDS patients were included in the analysis, of whom 179 (19.3%) developed AKI during hospitalization. A total of 43 features were used to build the model. Among all models, XGBoost performed the best. We used the top 10 features to build a compact model with an area under the curve (AUC) of 0.838, which improved to an AUC of 0.848 after the HPO. Conclusion Machine learning algorithms, especially XGBoost, are reliable tools for predicting AKI in ARDS patients. The compact model still retains excellent predictive ability, and the web-based calculator makes clinical usage more convenient.
Background: Early detection and intervention of disease deterioration are the keys to reducing the incidence of preventable intensive care unit cardiac arrest (ICU-CA). We aimed to investigate the ICU-CA predictive factors, including vital signs and laboratory indicators, and to analyze the performance of trends value of those factors on predicting ICU-CA. Methods: We conducted a matched case-control study at Qilu Hospital of Shandong University. Data on adult patients in ICU who suffered a cardiac arrest (CA) were retrospectively collected from 2016 to 2019, including vital signs and laboratory indicators at 48, 36, 24, 12, and 8 hours before ICU-CA. These cases were matched (ward, sex, and admission data) with controls (no ICU-CA) at a 1:2 ratio. Univariable logistic regression was used for statistical comparisons between cases and controls, and multivariate logistic regression was used to investigate the independent associations of indicators and their tendency with ICU-CA at given time points. The area under receiver operating characteristic (AUROC) was used to evaluate the predictive performance on ICU-CA.Results: Of 6164 ICU patients, 1042 patients suffered an ICU-CA during the 3 years. After careful screening, a total of 427 patients were included as the cases in the study, and 790 patients were included as controls. The vital signs and laboratory indicators at 8h before cardiac arrest, such as heart rate (HR), respiratory rate (RR), systolic blood pressure (SBP), oxygen saturation (SaO2), hemoglobin (HGB), potassium (K+), sodium (Na+), lactic acid (Lac), and pH all can predict the ICU-CA. The mean value, maximum value, minimum value, and range of these indicators were related to the occurrence of ICU-CA, and the trend values were more accurate than the current value for the variability in laboratory indicators. Conclusions: The ability of trends value of laboratory indicators for predicting ICU-CA was more accurate than the value at given time points for the variability in laboratory indicators. Adding trends of laboratory indicators may increase the accuracy of models designed to detect critical illness in ICU. Trial registration: ClinicalTrials.gov Identifier: NCT04670458.
scite is a Brooklyn-based organization that helps researchers better discover and understand research articles through Smart Citations–citations that display the context of the citation and describe whether the article provides supporting or contrasting evidence. scite is used by students and researchers from around the world and is funded in part by the National Science Foundation and the National Institute on Drug Abuse of the National Institutes of Health.
customersupport@researchsolutions.com
10624 S. Eastern Ave., Ste. A-614
Henderson, NV 89052, USA
This site is protected by reCAPTCHA and the Google Privacy Policy and Terms of Service apply.
Copyright © 2025 scite LLC. All rights reserved.
Made with 💙 for researchers
Part of the Research Solutions Family.