Artificial intelligence-assisted diagnosis and prognostication in low ejection fraction using electrocardiograms in inpatient department: a pragmatic randomized controlled trial
Open accessAbstract
Background: Early diagnosis of low ejection fraction (EF) remains challenging despite being a treatable condition. This study aimed to evaluate the effectiveness of an electrocardiogram (ECG)-based artificial intelligence (AI)-assisted clinical decision support tool in improving the early diagnosis of low EF among inpatient patients under non-cardiologist care. Methods: We conducted a pragmatic randomized controlled trial at an academic medical center in Taiwan. 13,631 inpatient patients were randomized to either the intervention group ( n = 6,840) receiving AI-generated ECG results or the control group ( n = 6,791) following standard care. The primary outcome was the incidence of newly diagnosed low EF (≤ 50%) within 30 days following the ECG. Secondary outcomes included echocardiogram utilization rates, positive predictive value for low EF detection, and cardiology consultation rates. Statistical analysis included hazard ratios (HR) with 95% confidence intervals (CI) for time-to-event outcomes and chi-square tests for categorical variables. Results: The intervention significantly increased the detection of newly diagnosed low EF in the overall cohort (1.5% vs. 1.1%, HR 1.50, 95% CI: 1.11–2.03, P = 0.023), with a more pronounced effect among AI-identified high-risk patients (13.0% vs. 8.9%, HR 1.55, 95% CI: 1.08–2.21). While overall echocardiogram utilization remained similar between groups (17.1% vs. 17.3%, HR 1.00, 95% CI: 0.92–1.09), the intervention group demonstrated higher positive predictive value for identifying low EF among patients receiving echocardiogram (34.2% vs. 20.2%, p < 0.001). Post-hoc analysis revealed increased cardiology consultation rates among high-risk patients in the intervention group (29.3% vs. 23.5%, p = 0.027). Conclusions: Implementation of an AI-ECG algorithm enhanced the early diagnosis of low EF in the inpatient setting, primarily by improving diagnostic efficiency rather than increasing overall healthcare utilization. The tool was particularly effective in identifying high-risk patients who benefited from increased specialist consultation and more targeted diagnostic testing. Trial registration: ClinicalTrials.gov Identifier: NCT05117970.
The claims and statistics of this paper haven’t been checked yet.
Run the check — free, no account needed. It reads the open-access text and checks the sources it cites.
Integrity signals
Registry
Cites
Statistics
Claims the paper makes
The claims and statistics of this paper haven’t been checked yet.
Run the check — free, no account needed. It reads the open-access text and checks the sources it cites.
Claims other papers make about this one
Each is checked against this paper’s own text: does it say what the citing paper says it does?
Low-LVEF detection is the most mature use case, having already progressed from model development to randomized trials and early implementation studies.
“Low-LVEF detection is the most mature use case, having already progressed from model development to randomized trials and early implementation studies.”
From AI-ECG for Echocardiography Triage in Structural Heart Disease: Evidence, Implementation, and Future Directions (2026) · cites this paper as [37]
The Methods describe a previously developed and validated low-LVEF model; the Background describes an earlier primary-care randomized trial, and the Discussion describes this pragmatic trial in everyday inpatient practice. These passages support progression from model development to trials and early clinical use, but they do not compare low-LVEF detection with other use cases or establish that it is the most mature. (LLM, conf=0.87)
Experimental — accuracy not yet measuredFollow-up imaging still occurred in only about half of AI-positive outpatients.
“Even so, follow-up imaging still occurred in only about half of AI-positive outpatients, and later inpatient pragmatic data likewise underscored a persistent gap between algorithmic flagging and downstream clinical action.14,37”
From AI-ECG for Echocardiography Triage in Structural Heart Disease: Evidence, Implementation, and Future Directions (2026) · cites this paper as [37]
The regulatory maturity of AI-ECG for SHD screening remains uneven.
“The regulatory maturity of AI-ECG for SHD screening remains uneven.”
From AI-ECG for Echocardiography Triage in Structural Heart Disease: Evidence, Implementation, and Future Directions (2026) · cites this paper as [37]
The Methods section mentions compliance with Taiwan Food and Drug Administration guidelines for this trial’s consent approach, but does not assess regulatory approval or maturity of AI-ECG tools. The Discussion addresses clinical validation and use across settings, not whether regulation of AI-ECG for structural heart disease screening is uneven. Searches for regulatory approval, clearance, SHD screening, and regulatory maturity found no passage addressing the claim. (LLM, conf=0.76)
Experimental — accuracy not yet measured
Self-check
Percentages, sums and reported differences re-computed from the paper’s own numbers.
The self-check needs the paper’s full text.
Cites
Heart Failure With Reduced Ejection Fraction: A Review
Murphy, S. P., Ibrahim, N. E., & Januzzi, J. L. (2020) · JAMA · doi:10.1001/jama.2020.10262 · Cited by 769 · see citing works
Natural history of asymptomatic left ventricular systolic dysfunction in the community
Wang, T. J., Evans, J. C., Benjamin, E. J., Levy, D., LeRoy, E. C., & Vasan, R. S. (2003) · Circulation · doi:10.1161/01.cir.0000085166.44904.79 · Cited by 633 · see citing works
The Epidemiology of Asymptomatic Left Ventricular Systolic Dysfunction: Implications for Screening
Wang, T. J., Levy, D., Benjamin, E. J., & Vasan, R. S. (2003) · Annals of Internal Medicine · doi:10.7326/0003-4819-138-11-200306030-00012 · Cited by 229 · see citing works
Brennan, M., Puri, S., Ozrazgat-Baslanti, T., Feng, Z., Ruppert, M., Hashemighouchani, H., Momcilovic, P., Li, X., Wang, D. Z., & Bihorac, A. (2019) · Surgery · doi:10.1016/j.surg.2019.01.002 · Cited by 131 · see citing works
Automation bias in electronic prescribing
Lyell, D., Magrabi, F., Raban, M. Z., Pont, L. G., Baysari, M. T., Day, R. O., & Coiera, E. (2017) · BMC Medical Informatics and Decision Making · doi:10.1186/s12911-017-0425-5 · Cited by 103 · cited 1× · see citing works
“Such discrepancies may elevate the risk to patient safety [25].”
Flagged first, then by citations
Research works printed but not linked: 21
- [1]Mozaffarian D, et al. Heart disease and stroke statistics–2015 update: a report from the American Heart Association. Circulation. 2015;131:e29-322. 10.1161/CIR.0000000000000152. 25520374 · DOI · Paper page
- [2]Ponikowski P, et al. 2016 ESC Guidelines for the diagnosis and treatment of acute and chronic heart failure: the task force for the diagnosis and treatment of acute and chronic heart failure of the European Society of Cardiology (ESC)Developed with the special contribution of the Heart Failure Association (HFA) of the ESC. Eur Heart J. 2016;37:2129–200. 10.1093/eurheartj/ehw128. 27206819 · DOI · Paper page
- [3]Yancy CW, et al. 2017 ACC/AHA/HFSA focused update of the 2013 ACCF/AHA guideline for the management of heart failure: a report of the American College of Cardiology/American Heart Association Task Force on clinical practice guidelines and the Heart Failure Society of America. Circulation. 2017;136:e137–61. 10.1161/CIR.0000000000000509. 28455343 · DOI · Paper page
- [4]Yusuf S, Pitt B, Davis CE, Hood WB Jr, Cohn JN. Effect of enalapril on mortality and the development of heart failure in asymptomatic patients with reduced left ventricular ejection fractions. N Engl J Med. 1992;327:685–91. 10.1056/NEJM199209033271003. 1463530 · DOI · Paper page
- [5]Pfeffer MA, et al. Effect of captopril on mortality and morbidity in patients with left ventricular dysfunction after myocardial infarction. Results of the survival and ventricular enlargement trial. The SAVE investigators. N Engl J Med. 1992;327:669–77. 10.1056/NEJM199209033271001. 1386652 · DOI · Paper page
- [6]Echouffo-Tcheugui JB, Erqou S, Butler J, Yancy CW, Fonarow GC. Assessing the risk of progression from asymptomatic left ventricular dysfunction to overt heart failure: a systematic overview and meta-analysis. JACC: Heart Fail. 2016;4(4):237–48. 26682794
- [7]Galasko GI, Barnes SC, Collinson P, Lahiri A, Senior R. What is the most cost-effective strategy to screen for left ventricular systolic dysfunction: natriuretic peptides, the electrocardiogram, hand-held echocardiography, traditional echocardiography, or their combination? Eur Heart J. 2006;27:193–200. 16267076 · DOI · Paper page
- [8]Nielsen O, et al. Value of BNP to estimate cardiac risk in patients on cardioactive treatment in primary care. Eur J Heart Fail. 2007;9:1178–85. 18062902 · DOI
- [9]Nielsen OW, McDonagh TA, Robb SD, Dargie HJ. Retrospective analysis of thecost-effectiveness of using plasmabrain natriuretic peptide inscreening for left ventricularsystolic dysfunction in the general population. J Am Coll Cardiol. 2003;41:113–20. 12570953 · DOI
- [10]Goudie BM, et al. Screening for left ventricular systolic dysfunction using GP-reported ECGs. Br J Gen Pract. 2007;57:191–5. 17359605 PMC2042566
- [11]Reichlin T, et al. Advanced ECG in 2016: is there more than just a tracing? Swiss Med Wkly. 2016;146:w14303–w14303. 27124801
- [12]Chen H-Y, et al. Artificial intelligence-enabled electrocardiography predicts left ventricular dysfunction and future cardiovascular outcomes: a retrospective analysis. J Pers Med. 2022;12:455. 35330455 PMC8950054 · DOI
- [13]Yao X, et al. Artificial intelligence–enabled electrocardiograms for identification of patients with low ejection fraction: a pragmatic, randomized clinical trial. Nat Med. 2021;27:815–9. 10.1038/s41591-021-01335-4. 33958795 · DOI · Paper page
- [14]Chen HY, et al. Artificial intelligence-enabled electrocardiogram predicted left ventricle diameter as an independent risk factor of long-term cardiovascular outcome in patients with normal ejection fraction. Front Med (Lausanne). 2022;9:870523. 10.3389/fmed.2022.870523. 35479951 PMC9035739 · DOI
- [16]Attia ZI, et al. Screening for cardiac contractile dysfunction using an artificial intelligence–enabled electrocardiogram. Nat Med. 2019;25:70–4. 30617318 · DOI · Paper page
- [17]Adedinsewo D, et al. Artificial intelligence-enabled ECG algorithm to identify patients with left ventricular systolic dysfunction presenting to the emergency department with dyspnea. Circ Arrhythm Electrophysiol. 2020;13:e008437. 32986471 · DOI · Paper page
- [18]Haq KT, Howell SJ, Tereshchenko LG. Applying artificial intelligence to ECG analysis: promise of a better future. Am Heart Assoc. 2020;13:e009111. 32809878
- [19]Kashou AH, et al. Artificial intelligence–augmented electrocardiogram detection of left ventricular systolic dysfunction in the general population. Mayo Clin Proc. Elsevier; 2021. PMC9904428 34120755
- [22]Mertens L, Friedberg MK. The gold standard for noninvasive imaging in congenital heart disease: echocardiography. Curr Opin Cardiol. 2009;24:119–24. 10.1097/HCO.0b013e328323d86f. 19225295 · DOI
- [23]van Baalen S, Boon M, Verhoef P. From clinical decision support to clinical reasoning support systems. J Eval Clin Pract. 2021;27:520–8. 33554432 PMC8248191 · DOI
- [26]Attia ZI, et al. Screening for cardiac contractile dysfunction using an artificial intelligence–enabled electrocardiogram. Nat Med. 2019;25:70–4. 10.1038/s41591-018-0240-2. 30617318 · DOI · Paper page
Cited by
Sorted by citations
Tang, Q., Deng, K., Cui, Y., Liu, H., & Qian, B. (2026) · International Journal of General Medicine · doi:10.2147/ijgm.s610298
- Authors
- Tsai, D.-J., Lin, C., Liu, W.-T., Lee, C.-C., Chang, C.-H., Lin, W.-Y., Liu, Y.-L., Chang, D.-W., Hsieh, P.-H., Tsai, C.-S., Chen, Y.-H., Hung, Y.-J., & Lin, C.-S.
- Affiliations
- Tri-Service General Hospital (TW)
- National Defense Medical Center (TW)
- Corresponding author
- Hung, Yi-Jen
- Lin, Chin-Sheng
- ORCID
- Venue
- BMC Medicine
- Published
- 2025-06-09
- Volume
- 23
- Issue
- 1
- Pages
- 342-342
- ISSN
- 1741-7015 (ISSN-L)
- PMC
- PMC12147261
- OpenAlex
- W4411127829
- Cited by
- 19
- References
- 5
- Citations by year
- 2026: 16 · 2025: 3
- FWCI
- 13.27
- Type
- journal article
- Language
- en
- Open access
- gold
- doi.org
- Locations
- BMC Medicine (publishedVersion, cc-by)
- PubMed Central (publishedVersion, cc-by)
- PubMed (publishedVersion)
- DOAJ (DOAJ: Directory of Open Access Journals) (submittedVersion, cc-by-sa)
- Funders
- National Science and Technology Council (NSTC 112-2222-E-030 -002 -MY2)
- Topics
- Health Sciences › Medicine › Cardiology and Cardiovascular Medicine › ECG Monitoring and Analysis
- Health Sciences › Medicine › Health Informatics › Artificial Intelligence in Healthcare and Education
- Health Sciences › Health Professions › Health Information Management › Artificial Intelligence in Healthcare
- Keywords
- reduced ejection fraction, artificial intelligence, electrocardiogram, clinical decision support, early diagnosis, hospitalized patients, positive predictive value, cardiology consultation, high-risk patients, heart failure, heart failure with reduced ejection fraction, heart failure with preserved ejection fraction
- MeSH terms
- Aged, Artificial Intelligence, Electrocardiography, Female, Humans, Inpatients, Male, Middle Aged, Prognosis, Stroke Volume, Taiwan
About these checks
Registry, retraction, Cites and statistics results are computed, not judged, with one exception: a printed reference that no index lists is read once by a model before it counts as a citation error. Claims are extracted by a model and kept only when their quote appears verbatim in the text; each cited source is then checked, with a model’s help, for whether it says what the claim relies on. Those verdicts can be wrong — read the claim’s quote and the cited source before relying on one.
Statistics and arithmetic are recomputed, not judged.
Confidence bands are shown only for fields calibrated on our annotated test set. None is calibrated yet, so no band is shown.
What counts as a claim ›How we measure confidence ›How we check this ›
Are you an author, or do you see a problem with this page? Request a review
