[Objective] To explore the risk factors affecting acute respiratory failure (ARF) after radical resection of esophageal cancer surgery and construct a Nomogram prediction model. [Methods] A retrospective analysis was conducted on 49 ARF patients and 97 non ARF patients after radical resection of esophageal cancer surgery in the First Affiliated Hospital of Hebei North University from December 2018 to December 2022. The patient clinical data was collected and compared to analyze the factors affecting ARF after radical resection of esophageal cancer surgery. The receiver operating characteristic (ROC) curves of the factors with statistically significant differences were drew. The area under the curve (AUC) and optimal cutoff value of each factor in predicting ARF after radical resection of esophageal cancer surgery were analyzed. Logistic regression model was used to analyze the independent risk factors for ARF after radical resection of esophageal cancer surgery. R language software 4.0 "rms" package was used to construct a Nomogram model for predicting ARF after radical resection of esophageal cancer surgery. The calibration curve and decision curve were used to internally validate the Nomogram model and evaluate its predictive performance. [Results] There were statistically significant differences in age, serum albumin level, surgery duration, smoking history, lung surgery history, anastomotic fistula and pleural adhesion between the two groups (P<0.05). AUC of age, serum albumin level and surgery duration were 0.761, 0.692 and 0.712, respectively. The optimal cutoff values were 54 years old, 38.15 g/L and 3.08 h, respectively. Age (>54 years old), serum albumin level (<38.15 g/L), surgery duration (>3.08 h), smoking history (yes), lung surgery history (yes), anastomotic fistula (yes) and pleural adhesion (yes) were the independent risk factors for ARF after radical resection of esophageal cancer surgery. C-index of the Nomogram model for predicting the affection of ARF after radical resection of esophageal cancer surgery was 0.725 (95% CI: 0.640-0.772), with a threshold greater than 0.21. The calibration curve showed good consistency between observed values and predicted values. The Nomogram model provided clinical net benefits that were higher than all the independent predictive factors. [Conclusion] The Nomogram model has good predictive performance and clinical application value, which can provide a certain reference for the prevention of ARF after radical resection of esophageal cancer surgery. |