BACKGROUND
Type 2 respiratory failure(T2RF) is one of the main causes of death in patients with acute exacerbations of chronic obstructive pulmonary disease (AECOPD), which has a rapid onset and serious consequences.
OBJECTIVE
This study aimed to identify the risk factors of AECOPD complicated with T2RF and establish a predictive model to provide early warnings for the disease.
METHODS
A total of 1252 patients over 40 years old with AECOPD were selected in 7 affiliated medical institutions of the Chongqing Medical University from January 1, 2014, to December 31, 2020. These patients were divided into two groups: the case group (n = 242) and the control group (n = 1010), according to the occurrence of T2RF during hospitalisation. Univariate analysis and the ‘least absolute shrinkage and selection operator (LASSO) model with multivariate logistic regression’ were used to screen independent risk factors. Three machine learning algorithms, random forest (RF), XGBoost and support vector machine (SVM), were used to construct the prediction model, and the logistic model was used as the control model. A nomogram was established based on the logistic model and proved to have good calibration and clinical applicability.
RESULTS
A total of 24 indicators were included in the study, among which 17 were selected by univariate analysis with statistically significant differences. Five independent risk factors were screened out by the ‘LASSO model with multivariate logistic regression’, including the duration of disease, procalcitonin, D-dimer, the neutrophil/lymphocyte ratio and pulmonary ventilation function. The area under the ROC curve of the logistic, RF, XGBoost and SVM models were 0.886, 0.903, 0.876 and 0.869, respectively, in the test set.
CONCLUSIONS
The clinical prediction model constructed in this study has a good predictive effect on AECOPD complicated with T2RF, and it can be used to predict the same in Chongqing.