Echocardiographic Evaluation in Children with Post-Acute Sequelae of SARS-CoV-2 Infection Using Deep Learning

Area of Science:

• Cardiology
• Artificial Intelligence
• Pediatrics

Background:

• Post-acute sequelae of SARS-CoV-2 infection (PASC) can lead to cardiac complications in children.
• Echocardiography is crucial for detecting cardiac abnormalities, but subtle changes may be missed by standard analysis.

Purpose of the Study:

• To apply deep learning to enhance the detection and understanding of echocardiographic changes in children with PASC.
• To evaluate the performance of an AI model in distinguishing cardiac function between children with PASC and controls.

Main Methods:

• A case-control study involving 270 children with PASC and 400 age-matched controls.
• Echocardiographic images were analyzed using a ResNet-50-based deep learning model.
• Exclusion criteria included congenital heart disease, inflammatory conditions, or arrhythmias.

Main Results:

• Standard echocardiographic parameters showed no significant abnormalities in the PASC group.
• The deep learning model achieved high accuracy (96.6%), sensitivity (96.7%), and specificity (96.2%).
• AI-assisted analysis effectively distinguished cardiac function between PASC and control groups.

Conclusions:

• Deep learning models can enhance the detection of subtle cardiac changes in children with PASC.
• AI-assisted echocardiography demonstrates high performance in identifying PASC-related cardiac differences.
• Further large-scale studies are needed to determine the clinical significance of these AI-detected differences.