Human vs Artificial Intelligence-Based Echocardiography Analysis as Predictor of Outcomes: An analysis from the World Alliance Societies of Echocardiography COVID study.

MedStar author(s):
Citation: Journal of the American Society of Echocardiography. 2022 Jul 18PMID: 35863542Institution: MedStar Health Research InstituteForm of publication: Journal ArticleMedline article type(s): Journal ArticleSubject headings: IN PROCESS -- NOT YET INDEXEDYear: 2022Local holdings: Available online through MWHC library: 2007 - presentISSN:
  • 0894-7317
Name of journal: Journal of the American Society of Echocardiography : official publication of the American Society of EchocardiographyAbstract: BACKGROUND: Transthoracic echocardiography (TTE) is the leading cardiac imaging modality for patients admitted with COVID-19 infection, a condition of high short-term mortality. We aimed to test the hypothesis that artificial intelligence (AI) based analysis of echocardiographic images could predict mortality more accurately than conventional analysis by a human expert.CONCLUSIONS: AI-based analysis of LVEF and LVLS had a similar feasibility to manual analysis, minimized variability and consequently increased the statistical power to predict mortality. AI-based analyses, but not manual, were significant predictors of in-hospital and follow-up mortality. Copyright © 2022. Published by Elsevier Inc.METHODS: Patients admitted to 13 hospitals for acute COVID-19 disease who had a TTE were included. Left ventricular (LV) ejection fraction (EF) and LV longitudinal strain (LS) were obtained manually by multiple expert readers and by an automated, AI software. The ability of the manual and AI analyses to predict all-cause mortality was compared.RESULTS: 870 patients were enrolled, mortality was 27.4% at a follow-up of 230+/-115 days. AI analysis had lower variability than manual for both LV EF (p=0.003) and LS (p=0.005). AI-derived LV EF and LS were predictors of mortality in univariable and multivariable regression analysis (OR=0.974, 95% CI= 0.956-0.991, p=0.003 for EF; OR=1.060, 95% CI 1.019-1.105, p=0.004 for LS), but LV EF and LS obtained by manual analysis were not. Direct comparison of predictive value of AI vs manual measurements of LV EF and LS was significantly better for AI (p=0.005 and 0.003 respectively). In addition, AI-derived LV EF and LS had more significant and stronger correlations to other objective biomarkers for acute disease than manual reads.All authors: Addetia K, Alizadehasl A, Asch FM, Citro R, Descamps T, Karagodin I, Lang RM, Monaghan MJ, Moreo A, Mostafavi A, Narang A, Ordonez Salazar BA, Sarwar R, Singulane CC, Soulat-Dufour L, Tucay ES, Tude Rodrigues AC, Upton R, Vasquez-Ortiz ZY, WASE-COVID Investigators, Woodward GM, Wu C, Xie MFiscal year: FY2023Digital Object Identifier: Date added to catalog: 2022-09-26
Holdings
Item type Current library Collection Call number Status Date due Barcode
Journal Article MedStar Authors Catalog Article 35863542 Available 35863542

Available online through MWHC library: 2007 - present

BACKGROUND: Transthoracic echocardiography (TTE) is the leading cardiac imaging modality for patients admitted with COVID-19 infection, a condition of high short-term mortality. We aimed to test the hypothesis that artificial intelligence (AI) based analysis of echocardiographic images could predict mortality more accurately than conventional analysis by a human expert.

CONCLUSIONS: AI-based analysis of LVEF and LVLS had a similar feasibility to manual analysis, minimized variability and consequently increased the statistical power to predict mortality. AI-based analyses, but not manual, were significant predictors of in-hospital and follow-up mortality. Copyright © 2022. Published by Elsevier Inc.

METHODS: Patients admitted to 13 hospitals for acute COVID-19 disease who had a TTE were included. Left ventricular (LV) ejection fraction (EF) and LV longitudinal strain (LS) were obtained manually by multiple expert readers and by an automated, AI software. The ability of the manual and AI analyses to predict all-cause mortality was compared.

RESULTS: 870 patients were enrolled, mortality was 27.4% at a follow-up of 230+/-115 days. AI analysis had lower variability than manual for both LV EF (p=0.003) and LS (p=0.005). AI-derived LV EF and LS were predictors of mortality in univariable and multivariable regression analysis (OR=0.974, 95% CI= 0.956-0.991, p=0.003 for EF; OR=1.060, 95% CI 1.019-1.105, p=0.004 for LS), but LV EF and LS obtained by manual analysis were not. Direct comparison of predictive value of AI vs manual measurements of LV EF and LS was significantly better for AI (p=0.005 and 0.003 respectively). In addition, AI-derived LV EF and LS had more significant and stronger correlations to other objective biomarkers for acute disease than manual reads.

English

Powered by Koha