Development and Validation of a Deep-Learning Model to Detect CRP Level from the Electrocardiogram

Area of Science:

• Cardiology
• Artificial Intelligence
• Biomarker Detection

Background:

• C-reactive protein (CRP) is a key inflammatory marker predicting arrhythmia occurrence and prognosis.
• The impact of inflammation on electrocardiographic (ECG) features remains largely unexplored.

Purpose of the Study:

• To assess a deep learning model's capability in detecting CRP levels using ECG data from patients in sinus rhythm.
• To investigate potential ECG alterations associated with elevated CRP levels.

Main Methods:

• Utilized 12,315 ECGs from 11,480 patients in sinus rhythm from a Guangzhou heart disease survey.
• Trained and validated a convolutional neural network to identify high CRP levels (>5mg/L) from 12-lead ECGs.
• Evaluated model performance using AUC, accuracy, sensitivity, specificity, and F1 score.

Main Results:

• High CRP group exhibited higher prevalence of overweight, smoking, hypertension, and diabetes (p < 0.05).
• ECG analysis revealed faster heart rate, longer QTc interval, and narrower QRS width in the high CRP group.
• The AI model achieved an AUC of 0.86 on the validation set and 0.85 on the testing set for CRP detection.

Conclusions:

• An AI-powered ECG algorithm effectively detects CRP levels in patients with sinus rhythm.
• Demonstrated that inflammation influences cardiac electrophysiological signals.
• Established a noninvasive screening approach for inflammatory status using ECG analysis.