Disease-focused evaluation of AI-ECG tools: clarifying study design, findings, and the principles of open science

Q: What is the clinical evidence from this study?

Study design: Letter. Population: Hypertrophic cardiomyopathy (HCM). Intervention: AI-ECG tools (AI-ECG HCM score, PRESENT-SHD) vs. None (case-only cohort).

Q: What does this research mean for the field?

AI-ECG probabilities in a confirmed hypertrophic cardiomyopathy cohort reflect phenotypic heterogeneity and track with markers of disease severity rather than providing binary diagnostic discrimination. Novelty: ClaimNovelty.METHODOLOGICAL. Consensus alignment: ConsensusAlignment.NEUTRAL.

Gamze Babür Güler; Arda Güler; Ibrahim Halil Tanboga

doi:10.1093/ehjdh/ztag120

July 24, 2026European Heart Journal - Digital HealthOpen Access

Disease-focused evaluation of AI-ECG tools: clarifying study design, findings, and the principles of open science

View Full Paper

Ask AI

Bookmark

Share

Key Points

To clarify the scope, methodology, and interpretation of disease-focused, case-only evaluations of AI-ECG tools in hypertrophic cardiomyopathy.
Evaluated AI-ECG score distributions within a cohort exclusively comprising patients with confirmed hypertrophic cardiomyopathy, including individuals with prior myectomy, alcohol septal ablation, or implantable cardioverter defibrillators.
Assessed associations between continuous AI model probability outputs (including PRESENT-SHD) and clinical, electrocardiographic, echocardiographic, and cardiac magnetic resonance markers of disease severity.
Hypertrophic cardiomyopathy AI model probabilities were associated with maximal wall thickness, NT-proBNP, left atrial diameter, Sokolow–Lyon index, T-wave axis, late gadolinium enhancement, and apical phenotype.
PRESENT-SHD model probabilities correlated with NT-proBNP, QTc interval, QRS duration, T-axis, left atrial diameter, LVEF, TAPSE, late gadolinium enhancement, diffuse LGE, atrial fibrillation, and bundle branch block.
AI-derived probability outputs exhibited broad variation across phenotypic subsets and were markedly lower in the subgroup presenting with normal ECGs.

Structured PICO

P

Population

A large case-only cohort of patients with confirmed hypertrophic cardiomyopathy used to characterize the outputs of AI-ECG tools.

E

Exposure

AI-ECG tools (including PRESENT-SHD)

O

Outcome

Association of AI-ECG generated outputs with clinical, electrocardiographic, echocardiographic, and cardiac magnetic resonance markers of disease expressionsurrogate

Limitations

Case-only design prevents estimation of formal diagnostic discrimination (AUROC, sensitivity, specificity) for HCM detection against a non-HCM comparator group.
Prior interventions such as myectomy, alcohol septal ablation, or ICD implantation may influence ECG morphology and model outputs.
Case-only design cannot estimate formal diagnostic discrimination or threshold-based diagnostic performance for HCM detection against non-cases.
Prior interventions (e.g., myectomy, septal ablation, ICD) and acquisition details may influence ECG morphology and model outputs.

Implication

AI-ECG tools can capture gradients of phenotypic expression and structural/functional disease burden within a confirmed HCM cohort, demonstrating utility beyond binary diagnostic classification.

Abstract

We thank Dhingra et al . for their thoughtful engagement with our study and, most importantly, for making these AI-ECG tools freely available to the research community. 1–5 We agree that independent evaluations should clearly distinguish the intended use of a model, the study population, the operating thresholds

Discussion

No takes yet. Share an insight, caveat, or question.

Cite This Study

Güler et al. (2026) conducted a letter in Hypertrophic cardiomyopathy (HCM). AI-ECG tools (AI-ECG HCM score, PRESENT-SHD) vs. None (case-only cohort) was evaluated.

synapsesocial.com/papers/6ad115de00741cc3fba7dae4 https://doi.org/10.1093/ehjdh/ztag120

View Full Paper

Ask AI

Bookmark

Share

Disease-focused evaluation of AI-ECG tools: clarifying study design, findings, and the principles of open science

Key Points

Structured PICO

Limitations

Implication

Abstract

Discussion

Cite This Study

Disease-focused evaluation of AI-ECG tools: clarifying study design, findings, and the principles of open science

Key Points

Structured PICO

Limitations

Implication

Abstract

Discussion

Cite This Study

Keep reading

More related papers

More from European Heart Journal - Digital Health

Keep reading

More related papers

More from European Heart Journal - Digital Health