Electrocardiogram-Artificial Intelligence and Immune-Mediated Necrotizing Myopathy: Predicting Left Ventricular...

Area of Science:

• Cardiology
• Neurology
• Artificial Intelligence

Background:

• Immune-mediated necrotizing myopathy (IMNM) can lead to cardiac complications, including left ventricular dysfunction (LVD).
• Early detection of LVD is crucial for managing IMNM and improving patient outcomes.
• Existing diagnostic methods for LVD may have limitations in accessibility or timeliness.

Purpose of the Study:

• To evaluate the effectiveness of an existing electrocardiogram (ECG)-artificial intelligence (AI) algorithm in detecting LVD in patients with IMNM.
• To assess the algorithm's accuracy in predicting LVD compared to echocardiography.
• To determine the algorithm's utility in monitoring treatment response and predicting mortality risk in IMNM.

Main Methods:

• A retrospective observational study of IMNM patients diagnosed between 2000 and 2020.
• Application of a validated AI algorithm to standard 12-lead ECGs to predict LVD probability.
• Comparison of AI-derived LVD probability scores with echocardiogram findings, immunotherapy response, and mortality data.

Main Results:

• The ECG-AI algorithm demonstrated acceptable accuracy (74%) in predicting LVD in IMNM patients.
• The algorithm showed a sensitivity of 80.0% and specificity of 62.8% for LVD detection.
• Abnormal LVD probability scores were associated with a 7-fold increased mortality risk and predicted early LVD in asymptomatic patients.

Conclusions:

• An AI-ECG algorithm can effectively assist in detecting LVD in IMNM patients.
• The algorithm aids in decisions regarding further echocardiogram testing and immunotherapy management.
• AI-ECG analysis provides valuable information on mortality risk, crucial for patient monitoring and treatment escalation in IMNM.