Recurrent Hemoptysis After Bronchial Artery Embolization: Prediction Using a Nomogram and Artificial Neural Network...

Area of Science:

• Interventional Radiology
• Pulmonary Medicine
• Predictive Analytics

Background:

• Recurrent hemoptysis poses a significant risk following bronchial artery embolization (BAE).
• Effective prediction of recurrent bleeding is crucial for patient management and treatment planning.

Purpose of the Study:

• To develop and validate a nomogram and artificial neural network (ANN) model for predicting recurrent hemoptysis post-BAE.
• To identify key predictors associated with hemoptysis recurrence.

Main Methods:

• A retrospective analysis of 242 patients undergoing BAE was conducted, split into training (141) and validation (101) cohorts.
• Univariable and multivariable analyses identified predictors, including age, lung cancer, bronchial-pulmonary shunts, and nonbronchial systemic artery involvement.
• A nomogram and ANN model were constructed and validated using Harrell C statistics and ROC curves.

Main Results:

• Older age (≥60), lung cancer, bronchial-pulmonary shunts, and nonbronchial systemic artery involvement were significant predictors of recurrent hemoptysis.
• The developed nomogram and ANN model demonstrated high predictive accuracy, with Harrell C statistics of 0.849 (internal) and 0.799 (external).
• An optimal risk cutoff of 0.16 was identified for recurrent hemoptysis.

Conclusions:

• The nomogram and ANN model provide an effective tool for predicting recurrent hemoptysis risk after BAE.
• Patients identified with a high risk (> 0.16) may benefit from further interventions.
• These predictive models can enhance personalized treatment strategies for hemoptysis patients.