This research is Trending!
AI-ECG Detection of Structural Heart Disease in the Community Setting
Timothy J. PoteruchaJ. Weston HughesMichael I. BrenerMichael L. ChuangNadira HamidClayton SnyderJulia ThompsonJena UngartenKayo TakahashiBrian R. LindmanFrancisco Alberto Roedan OliverJoshua FinerDustin HartzelHeidi HartmanAaron LongDaniel RochaPaloma MaltaMichelle CastilloRichard RicartJeffrey RuhlDavid vanMaanenAlbert J. RogersMarco PerezOlivier TastetRobert AvramRaphael NishimuraJuan F. GranadaMichael J. MackMartin B. LeonPierre EliasDavid J. Cohen
Key result
EchoNext AI-ECG model shows lower structural heart disease discrimination in community versus hospital cohorts.
- n=2,402
Why the study?
It was unknown how differences in disease spectrum influence the transportability of AI-based diagnostic models developed in hospital populations when deployed into community-dwelling populations.
Does the EchoNext AI-ECG model accurately detect structural heart disease in a community-dwelling population compared to hospital-based cohorts?
Key Points
- To evaluate how differences in disease spectrum influence the transportability and diagnostic performance of an AI-ECG model when deployed in a community-dwelling population compared to hospital cohorts.
- Evaluated the EchoNext AI-ECG model in an interventional trial cohort (PREVUE-VALVE; NCT05357404) comprising community-dwelling individuals aged 65 to 85 years undergoing in-home ECG and transthoracic echocardiography (N=2,402 analyzed out of 3,000).
- Assessed diagnostic performance using the area under the receiver operating characteristic curve (AUC) and compared results to derivation and external hospital cohorts using propensity score matching to account for prevalence and case mix.
- The community cohort demonstrated a significantly lower prevalence of structural heart disease than the hospital-based cohort (8% vs 43%), presenting milder phenotypes with more moderate tricuspid regurgitation and less systolic heart failure.
- Model discrimination was lower in the community cohort than in the derivation hospital cohort (AUC: 71% [95% CI: 66%-76%] vs 83% [95% CI: 82%-83%]), a difference attenuated but not eliminated by propensity matching.
- Performance improved in community subgroups with greater disease severity, including individuals with an abnormal baseline ECG (AUC: 79% [95% CI: 75%-83%]) and impaired health status (AUC: 76% [95% CI: 70%-82%]).
Study Design
Type
Observational (n=2,402)
Multicenter
Yes
Structured PICO
Does the EchoNext AI-ECG model accurately detect structural heart disease in a community-dwelling population compared to hospital-based cohorts?
P
Population
2,402 community-dwelling individuals aged 65 to 85 years evaluated for structural heart disease using in-home ECG and echocardiography.
E
Exposure
EchoNext AI-ECG model for detecting structural heart disease
C
Comparator
Derivation and external hospital-based cohorts
O
Outcome
Model discrimination for structural heart disease assessed by area under the receiver operating characteristic curve (AUC)surrogate
Main Result
Absolute Event Rate: 71% vs 83%
AI-ECG models trained in hospital cohorts have attenuated diagnostic performance when applied to community-dwelling populations due to differences in disease prevalence and severity.
Ask about this paper
Answers are grounded in this paper and cite their sources.
Key expert perspectives
Captured external expert commentary on this paper, strongest first. Original sources are linked where available.
“These findings challenge the assumption of transportability of AI-based diagnostics across populations and emphasize the importance of AI model validation within intended use populations.”
Implication
AI-ECG performance declines outside hospitals; leaves open transportability to community settings and requires prospective validation.
Abstract
BACKGROUND: Artificial intelligence (AI) diagnostic models are typically developed in hospital-based populations enriched for disease prevalence and severity, where distinctions between health and disease are pronounced. When deployed in community-dwelling populations with lower prevalence and milder phenotypes, these distinctions may become less well-defined, raising questions about transportability and the impact of disease spectrum on model performance. OBJECTIVES: The purpose of this study was to evaluate how differences in disease spectrum influence the transportability of AI-based diagnostic models into community-dwelling populations, using artificial intelligence electrocardiogram analysis (AI-ECG) for structural heart disease (SHD) as a test case. METHODS: EchoNext is an AI-ECG model trained in a multicenter, hospital-based cohort to detect SHD. We evaluated its performance in PREVUE-VALVE (Age and Sex-Specific PREValence of AcqUirEd VALVular Heart DiseasE Study), a community-based study of individuals aged 65 to 85 years undergoing in-home ECG and transthoracic echocardiography. Performance was assessed using area under the receiver operating characteristic curve (AUC) and predictive value and compared with derivation and external hospital-based cohorts, using propensity matching to account for differences in disease prevalence and case mix. Subgroup analyses were performed in clinically relevant populations. RESULTS: Among 3,000 PREVUE-VALVE participants, 2,402 met criteria for analysis. Compared with hospital-based cohorts, PREVUE-VALVE had lower SHD prevalence (8% vs 43%), less severe disease, and a shift in phenotype, including more moderate tricuspid regurgitation and less systolic heart failure. Consistent with these differences, model discrimination was lower in PREVUE-VALVE than in the hospital-based cohort (AUC: 71% [95% CI: 66%-76%] vs 83% [95% CI: 82%-83%]). Propensity matching attenuated but did not eliminate this difference, while performance was similar across external hospital-based cohorts, supporting disease spectrum and clinical context as key drivers. Performance was modestly better in PREVUE-VALVE subgroups with higher SHD prevalence and greater disease severity, such as individuals with an abnormal ECG (AUC: 79% [95% CI: 75%-83%]) or impaired health status (AUC: 76% [95% CI: 70%-82%]). CONCLUSIONS: In a community-dwelling population with lower disease prevalence and milder phenotypes, AI-ECG performance was attenuated relative to hospital-based cohorts, driven by differences in disease spectrum. These findings underscore the impact of both disease prevalence and case mix on AI performance and highlight the importance of model evaluation within intended use populations. (Age and Sex-Specific PREValence if AcqUirEd VALVular Heart DiseasE; NCT05357404).
Discussion
No takes yet. Share an insight, caveat, or question.
Cite This Study
Poterucha et al. (2026) conducted an observational in Structural heart disease (n=2,402). EchoNext AI-ECG model vs. Hospital-based cohorts was evaluated on Model discrimination for structural heart disease (AUC). The EchoNext AI-ECG model demonstrated lower discrimination for structural heart disease in a community-dwelling population compared with hospital-based cohorts (AUC 71% vs 83%).