Validation of an Artificial Intelligence-Derived ECG Algorithm for Detecting Cardiac Amyloidosis in Patients With...

Area of Science:

• Cardiology
• Medical Diagnostics
• Artificial Intelligence in Medicine

Background:

• Cardiac transthyretin amyloidosis (ATTR-CA) is often underdiagnosed and presents as heart failure with preserved ejection fraction (HFpEF).
• Early diagnosis of ATTR-CA is crucial for initiating disease-modifying therapies.
• Current diagnostic challenges necessitate simple, accessible screening tools for routine clinical practice.

Purpose of the Study:

• To validate an artificial intelligence-derived, visually interpretable ECG algorithm for screening ATTR-CA in HFpEF patients.
• To assess the algorithm's accuracy, sensitivity, and specificity in both internal and external validation cohorts.
• To determine the association of the identified ECG pattern with ATTR-CA and patient survival.

Main Methods:

• A multicenter study involving 885 HF patients from two European centers.
• Internal validation used 560 patients (149 ATTR-CA, 318 HFpEF, 93 hypertrophic cardiomyopathy) with analyzable ECGs.
• External validation included 107 patients; ECGs were analyzed blindly using a 2-step AI-derived algorithm.

Main Results:

• The specific ECG pattern was detected in 82.6% of ATTR-CA patients versus 10.2% with HFpEF and 6.5% with hypertrophic cardiomyopathy (P<0.001).
• High accuracy was observed: internal AUC 0.87, sensitivity 83%, specificity 91%; external AUC 0.84, sensitivity 89%, specificity 79%.
• The pattern strongly correlated with ATTR-CA (OR 46; P<0.001) and was linked to reduced 3-year survival (P=0.007).

Conclusions:

• A visually interpretable, AI-derived ECG algorithm effectively screens for ATTR-CA in HFpEF patients.
• The algorithm's simplicity and compatibility with standard ECG systems facilitate broad clinical implementation.
• This tool supports earlier identification of ATTR-CA, enabling timely therapeutic interventions.