Predicting in-hospital rupture of type A aortic dissection using Random Forest

Area of Science:

• Cardiovascular Surgery
• Medical Prediction Modeling
• Aortic Disease Research

Background:

• Type A aortic dissection (TAAD) poses a significant risk of in-hospital rupture.
• Accurate prediction of rupture is crucial for timely surgical intervention and patient management.
• Existing tools may not adequately stratify rupture risk in TAAD patients.

Purpose of the Study:

• To develop and validate a prediction tool for in-hospital rupture in patients with type A aortic dissection (TAAD).
• To guide emergency surgical triage and improve patient counseling for TAAD.
• To identify key clinical and imaging predictors of TAAD rupture.

Main Methods:

• Retrospective analysis of 1,133 TAAD patients from January 2010 to December 2016.
• Data split into training (70%) and testing (30%) datasets.
• Random Forest classification model employed for prediction tool development.

Main Results:

• A Random Forest model identified 16 important predictors, including age, BMI, syncope, periaortic hematoma, and hemopericardium.
• The model demonstrated excellent discrimination in the training set (AUC 0.994).
• Validation in the testing set showed good predictive performance (AUC 0.752).

Conclusions:

• An accessible prediction tool for TAAD in-hospital rupture was successfully developed and validated.
• Periaortic hematoma emerged as the most significant predictor of rupture.
• Clinical factors like syncope are valuable for risk stratification in TAAD patients.