Transparent and robust Artificial intelligence-driven Electrocardiogram model for Left Ventricular Systolic Dysfunction

Heart failure (HF) is an escalating global health concern, worsened by an aging population and limitations in traditional diagnostic methods like electrocardiograms (ECG). The advent of deep learning has shown promise for utilizing 12-lead ECG models for the early detection of left ventricular systolic dysfunction (LVSD), a crucial HF indicator. This study validates the AiTiALVSD, an AI/machine learning-enabled Software as a Medical Device, for its effectiveness, transparency, and robustness in detecting LVSD. Conducted at Mediplex Sejong Hospital in the Republic of Korea, this retrospective single-center cohort study involved patients suspected of LVSD. The AiTiALVSD model, which is based on a deep learning algorithm, was assessed against echocardiography findings. To improve model transparency, the study utilized Testing with Concept Activation Vectors (TCAV) and included clustering analysis and robustness tests against ECG noise and lead reversals. The study involved 688 participants and found AiTiALVSD to have a high diagnostic performance, with an AUROC of 0.919. There was a significant correlation between AiTiALVSD scores and left ventricular ejection fraction values, confirming the model’s predictive accuracy. TCAV analysis showed the model’s alignment with medical knowledge, establishing its clinical plausibility. Despite its robustness to ECG artifacts, there was a noted decrease in specificity in the presence of ECG noise. AiTiALVSD’s high diagnostic accuracy, transparency, and resilience to common ECG discrepancies underscore its potential for early LVSD detection in clinical settings. This study highlights the importance of transparency and robustness in AI/ML-based diagnostics, setting a new benchmark in cardiac care.