Abstract
OBJECTIVE: To analyze the importance ranking of influencing factors of postoperative complications of free flap reconstruction in patients with head and neck cancer by using Logistic regression and random forest algorithm, and to construct and verify the prediction model. METHODS: The research subjects were patients with head and neck tumors who underwent free flap reconstruction in our hospital, The clinical-relevant data of all patients were collected. Patients were randomly divided into training set and validation set at the ratio of 7:3. Univariate and multivariate analyses using Logistic regression were performed to identify independent risk factors for postoperative complications. The random forest algorithm was further used to construct the prediction model, and the performance of the model was verified by receiver operating characteristic curve (ROC) analysis, calibration curve evaluation, and decision curve analysis (DCA). RESULTS: A total of 341 patients were included in the study, and 82 cases (24.05%) had postoperative complications. Multivariate Logistic regression analysis showed that age, hypertension, operation time, bleeding volume, flap type and flap area were the independent risk factors for complications after free flap reconstruction in patients with head and neck cancer (P < 0.05). The contribution magnitudes of each variable obtained from the random forest model was flap resection area, intraoperative bleeding volume, age, operation time, flap type and concomitant hypertension. The calibration curve of the constructed Nomogram model showed that the predicted value was in good agreement with the actual value, and the AUC of the ROC curve was 0.793 (95% CI: 0.711-0.874) and 0.788 (95% CI: 0.665-0.912), respectively, showing good prediction performance. The DCA analysis indicated that the model had good clinical application value. CONCLUSION: Logistic regression and random forest algorithm can effectively analyze the influencing factors of complications after free flap reconstruction in patients with head and neck cancer and construct accurate prediction model. This model can provide a scientific theoretical reference for the prevention of postoperative complications and facilitate the precise optimization of individualized prevention plans based on risk prediction.