EUVIMEDEUVIMED entdecken
Systematische ÜbersichtMetaanalyse

Abstract 8167: Artificial Intelligence-Enabled Single-Lead Electrocardiography for Structural Heart Disease Screening: A Systematic Review and Meta-Analysis

Abstract 8167: Künstlich-intelligenzgestützte Ein-Kanal-Elektrokardiographie zum Screening struktureller Herzerkrankungen: systematische Übersicht und Metaanalyse

Yassein Elhussein, Fatima Alzhara Abdalraheem, Mohamed Akram Abdelwahab Hassa Khalil, Abdelwahab Ahmed, Ahmed A Elhadi, Mustafa Yousif, Waheed Qaisi, Mubarak I. Abdalla, Shafaq Jabeen, Shehab Hashem Elsoghier, Osama Elbahy, Mohamed A Awad Elkarim und 1 weitere

Circulation · 02.11.2026 · 154 (Suppl_1)

DOI: 10.1161/circ.154.suppl_1.8167 ↗

Auf einen Blick

19 Studien untersuchten KI-gestützte Ein-Kanal-EKGs zum Screening struktureller Herzerkrankungen. Für LVSD zeigte die Metaanalyse eine gepoolte Sensitivität von 0,833 und Spezifität von 0,858.

Fragestellung

Wie gut erkennt KI-gestützte Ein-Kanal-Elektrokardiographie strukturelle Herzerkrankungen, insbesondere LVSD?

Methode

Systematische Übersicht und Metaanalyse; Suche in MEDLINE/PubMed, Embase, Web of Science, Scopus, Cochrane Library und IEEE Xplore bis Mai 2026; random-effects diagnostische Metaanalyse.

Ergebnisse

6 Studien mit 7.329 Teilnehmenden und 496 LVSD-Fällen in der Primäranalyse. Sensitivität 0,833 (95 %-KI 0,693–0,916), Spezifität 0,858 (95 %-KI 0,715–0,936), LR+ 5,903, LR- 0,201, DOR 31,148.

Einschränkungen

Substanziell hohe Heterogenität (I2=86,9 % für Sensitivität; I2=99,2 % für Spezifität).

Schlussfolgerung der Autor:innen

KI-gestützte Ein-Kanal-EKGs zeigen vielversprechende Leistung für das LVSD-Screening über verschiedene Plattformen hinweg; wegen der Heterogenität ist die Interpretation vorsichtig zu erfolgen.

Kernaussagen

  • LVSD-Screening: Sensitivität 0,833, Spezifität 0,858
  • 19 Studien insgesamt; Primäranalyse mit 6 Studien
  • Hohe Heterogenität zwischen Plattformen und Schwellenwerten

Deutsche Zusammenfassung: KI-gestützt aus dem Original-Abstract erstellt; alle Zahlen wurden automatisch mit dem Original abgeglichen. Maßgeblich ist der Original-Abstract.

Kurz gefasst

Systematische Übersicht in der Zeitschrift Circulation, erschienen am 02.11.2026. Verfasst von Yassein Elhussein und 12 weiteren Autor:innen.

Auf einen BlickAbstractBibliografische AngabenZitierenVerwandte Studien

Abstract (Originalsprache)

Introduction

Artificial intelligence (AI)-enabled single-lead electrocardiography may enable low-cost, scalable structural heart disease screening beyond imaging-based settings, but performance across wearable, portable, handheld, and ECG-enabled stethoscope platforms remains uncertain.

Methods

We searched MEDLINE/PubMed, Embase, Web of Science, Scopus, Cochrane Library, and IEEE Xplore through May 2026 for adult studies evaluating AI applied to single-lead wearable, handheld, portable, or ECG-enabled stethoscope ECG for structural heart disease detection. Eligible reference standards included echocardiography-based structural diagnosis and other accepted imaging-based confirmation where applicable. Primary synthesis focused on left ventricular systolic dysfunction (LVSD) studies with extractable 2x2 data. Random-effects diagnostic meta-analysis was performed for sensitivity, specificity, likelihood ratios, and diagnostic odds ratio. AUROC-only evidence and non-LVSD targets were summarized descriptively.

Results

Nineteen studies were included. The primary LVSD meta-analysis included six peer-reviewed full-text studies with 7,329 participants, 496 LVSD cases, and 6,833 non-LVSD controls. Acquisition approaches included smartwatch ECG, portable single-channel ECG, ECG-enabled stethoscope ECG, and derived/adapted single-lead ECG. Pooled sensitivity was 0.833 (95% CI, 0.693-0.916), and pooled specificity was 0.858 (95% CI, 0.715-0.936). Pooled positive likelihood ratio (LR+) was 5.903 (95% CI, 2.499-13.944), negative likelihood ratio (LR-) was 0.201 (95% CI, 0.104-0.388), and diagnostic odds ratio was 31.148 (95% CI, 7.674-126.426). Heterogeneity was substantial (I2=86.9% for sensitivity; I2=99.2% for specificity). Leave-one-out analysis identified the highest-specificity study as the strongest heterogeneity driver; after exclusion, diagnostic performance remained directionally consistent: sensitivity 0.805, specificity 0.797, LR+ 3.811, LR- 0.241, and DOR 16.069. AUROC-only LVSD evidence included 35 rows from 12 studies, with median AUROC 0.881 (range, 0.650-0.972). Evidence for LVH, valvular disease, and composite structural heart disease was limited.

Conclusion

AI-enabled single-lead ECG shows promising performance for LVSD screening across portable, wearable, and ECG-enabled stethoscope platforms. Substantial heterogeneity across acquisition methods, thresholds, validation settings, and populations supports cautious interpretation and highlights the need for standardized reporting, prospective external validation, and real-world implementation studies.

Zitieren

APA

Elhussein, Y., Abdalraheem, F. A., Khalil, M. A. A. H., Ahmed, A., Elhadi, A. A., Yousif, M., Qaisi, W., Abdalla, M. I., Jabeen, S., Elsoghier, S. H., Elbahy, O., Elkarim, M. A. A., & Elhussain, M. (2026). Abstract 8167: Artificial Intelligence-Enabled Single-Lead Electrocardiography for Structural Heart Disease Screening: A Systematic Review and Meta-Analysis. Circulation, 154(Suppl_1). https://doi.org/10.1161/circ.154.suppl_1.8167

Vancouver

Elhussein Y, Abdalraheem FA, Khalil MAAH, Ahmed A, Elhadi AA, Yousif M, et al. Abstract 8167: Artificial Intelligence-Enabled Single-Lead Electrocardiography for Structural Heart Disease Screening: A Systematic Review and Meta-Analysis. Circulation. 2026;154(Suppl_1). doi:10.1161/circ.154.suppl_1.8167

BibTeX
@article{Elhussein2026,
  author = {Elhussein, Yassein and Abdalraheem, Fatima Alzhara and Khalil, Mohamed Akram Abdelwahab Hassa and Ahmed, Abdelwahab and Elhadi, Ahmed A and Yousif, Mustafa and Qaisi, Waheed and Abdalla, Mubarak I. and Jabeen, Shafaq and Elsoghier, Shehab Hashem and Elbahy, Osama and Elkarim, Mohamed A Awad and Elhussain, Mohamed},
  title = {Abstract 8167: Artificial Intelligence-Enabled Single-Lead Electrocardiography for Structural Heart Disease Screening: A Systematic Review and Meta-Analysis},
  journal = {Circulation},
  year = {2026},
  volume = {154},
  number = {Suppl_1},
  doi = {10.1161/circ.154.suppl_1.8167}
}
Kostenlos starten →