An artificial intelligence based abdominal aortic aneurysm prognosis classifier to predict patient outcomes

Area of Science:

• Cardiovascular Medicine
• Biomedical Engineering
• Artificial Intelligence in Healthcare

Background:

• Abdominal aortic aneurysms (AAA) rupture is a leading cause of death, with current management relying on a "maximum diameter criterion."
• This criterion is insufficient, as smaller AAAs can rupture, leading to suboptimal patient outcomes.
• Accurate risk stratification for AAA is crucial for timely intervention and improved patient survival.

Purpose of the Study:

• To develop and validate a machine learning model for predicting AAA patient outcomes.
• To create an aneurysm prognosis classifier predicting stability, need for repair, or rupture.
• To offer a personalized approach to AAA management beyond the current diameter-based guidelines.

Main Methods:

• Trained and assessed machine learning models using clinical, biomechanical, and morphological data from 381 AAA patients.
• Utilized patient-specific medical imaging and clinical data as input for the models.
• Developed an Aneurysm Prognosis Classifier (APC) model to predict one of three patient outcomes.

Main Results:

• The study represents the largest cohort using medical imaging and clinical data for AAA outcome classification.
• The developed APC model demonstrates potential for stratifying specific patient outcomes.
• Machine learning models integrated patient-specific biomechanical, morphological, and clinical data.

Conclusions:

• The APC model shows promise as a clinical tool for personalized AAA management.
• This approach can assist clinicians in making more informed decisions for AAA patients.
• Integrating diverse patient data improves the accuracy of AAA rupture risk prediction.