Diagnostic and Prognostic Electrocardiogram-Based Models for Rapid Clinical Applications.

Area of Science:

• Cardiology
• Artificial Intelligence
• Medical Diagnostics

Background:

• Artificial intelligence (AI) is increasingly utilized for electrocardiogram (ECG) analysis, offering potential for improved diagnosis and prognosis of cardiovascular and noncardiac conditions.
• The past five years have seen significant advancements in AI-driven ECG applications, fueled by deep learning and wider adoption of ECG technologies.

Purpose of the Study:

• To review clinical studies and AI-enhanced ECG applications for cardiovascular disease detection, diagnosis, and prognosis from 2019-2023.
• To identify challenges and propose solutions for the effective clinical integration and global scalability of AI in ECG analysis.

Main Methods:

• Systematic review of clinical studies on AI-enhanced ECG analysis published between 2019 and 2023.
• Analysis of AI-ECG application development, validation, regulatory approval, and clinical implementation.

Main Results:

• The majority of published studies are single-center, retrospective, proof-of-concept investigations lacking external validation.
• Prospective studies suitable for clinical deployment constitute less than 15% of the research.
• FDA-approved AI-ECG applications often stem from commercial collaborations, with many targeting mobile or wearable devices.

Conclusions:

• The field of AI in ECG analysis is nascent, facing hurdles including prospective multicenter validation, technical issues, bias, data security, and generalizability.
• Overcoming these challenges is crucial for developing comprehensive, clinically integrated, and globally scalable AI solutions for cardiovascular disease management.