Application of artificial intelligence in carotid endarterectomy and carotid artery stenting: A systematic review

Abstract

Objectives

Carotid stenosis plays a significant role in stroke burden. Surgical intervention in the form of carotid endarterectomy or carotid artery stenting is an important stroke risk reduction strategy. Careful patient selection with identification of high-risk individuals is crucial to operative planning given perioperative risks including stroke, myocardial infarction, and death. Machine learning (ML) is a subset of artificial intelligence (AI) consisting of mathematical algorithms that can learn from datasets to perform particular tasks. These algorithms offer a tool for prediction of patient outcomes by analysis of preoperative data leading to improved patient selection. This systematic review aims to assess the use of artificial intelligence in risk stratification for carotid endarterectomy and carotid artery stenting.

Methods

PubMed, Web of Knowledge, EMBASE, and the Cochrane Library were systematically searched to identify any articles utilising artificial intelligence in predicting surgical outcomes in carotid endarterectomy or carotid artery stenting. After duplicate removal, all studies underwent independent title and abstract screening followed by quality assessment using the PROBAST tool. Data extraction was then carried out for synthesis and comparison of study outcomes including accuracy, area under receiver operator curve (AUC), sensitivity, and specificity.

Results

After duplicate processing, a total of 100 articles underwent title and abstract screening resulting in 11 clinical studies published between 2008 and 2023 that fit eligibility criteria. Surgical outcomes assessed included haemodynamic instability, shunt requirement, hyperperfusion syndrome, stroke, myocardial infarction, and death. Artificial intelligence models were able to accurately predict major adverse cardiovascular events (AUC 0.84), postoperative haemodynamic instability (AUC 0.86), shunt requirement (AUC 0.87), and postoperative hyperperfusion syndrome (AUC 0.95). However, many studies had a high risk of bias due to lack of external validation.

Conclusion

This systematic review highlights the potential application of machine learning in prediction of surgical outcomes in carotid artery intervention. However, use of these tools in a clinical setting requires further robust study with use of external validation and larger patient datasets.