Artificial intelligence-driven diagnosis of acute thoracic aortic dissection: integrating imaging, biomarkers, and clinical workflows-a narrative review.

Abstract

Background and objective: Patients presenting to the emergency department with acute thoracic aortic dissection (ATAD) often experience chest pain that requires urgent intervention. However, other chest pain-related emergencies, such as acute coronary syndrome (ACS) and acute pulmonary embolism (PE), are far more common and frequently overshadow ATAD. This disparity leads to a high rate of ATAD misdiagnosis. Recent advancements in artificial intelligence (AI) have led to the development of various models utilizing imaging modalities and biomarkers to enable rapid triage and diagnosis of ATAD in emergency settings. This article aims to evaluate the performance and clinical significance of these AI models within the context of clinical workflows.

Methods: We performed literature searches in PubMed, Scopus, and Web of Science to identify relevant studies published between 2015 and 2025, with the focus of the differentiation of ATAD patients from other chest pain-related conditions in emergency settings, with the application of AI.

Key content and findings: Eighteen studies were retrieved from the past ten years, highlighting a significant knowledge gap in the field of translational medicine. The discussion included an overview of AI-powered models for ATAD diagnosis, as well as guidelines on current clinical workflows and the application of AI in clinical settings.

Conclusions: This article offers a detailed review of AI models developed for the screening and diagnosis of ATAD. It highlights not only the performance of these technologies but also their clinical importance in facilitating timely interventions for high-risk patients. Looking forward, we anticipate a future where AI and deep learning (DL)-driven ATAD diagnostic models will play a pivotal role in optimizing ATAD clinical management.