Exploring artificial intelligence methods for cardiac syncope diagnosis combined with electrocardiogram parameters...

Area of Science:

• Cardiology
• Artificial Intelligence
• Medical Diagnostics

Background:

• Cardiac syncope poses a significant life-threatening risk.
• Current clinical screening tools for cardiac syncope are lacking.
• This study addresses the need for effective diagnostic methods.

Purpose of the Study:

• To develop and optimize artificial intelligence (AI) methods for automatic cardiac syncope diagnosis.
• To leverage electrocardiogram (ECG) parameters and clinical characteristics for improved diagnostic accuracy.
• To create a practical screening tool for cardiac syncope.

Main Methods:

• Utilized data from 380 syncope patients hospitalized between June 2018 and August 2022.
• Developed and compared six supervised machine learning models.
• Selected the top twelve predictive features based on importance ranking.
• Evaluated model performance using the area under the curve (AUC).

Main Results:

• The random forest model achieved the highest performance.
• Achieved an AUC of 0.85 in the development cohort and 0.75 in the validation cohort.
• The best model demonstrated sensitivity of 0.72 and specificity of 0.85.

Conclusions:

• AI methods show promise in classifying syncope.
• AI can potentially serve as a cost-effective and efficient screening tool for cardiac syncope.
• The developed AI approach has been proposed as a web service for practical application.