Human versus Artificial Intelligence–Based Echocardiographic Analysis as a Predictor of Outcomes: An Analysis from the World Alliance Societies of Echocardiography COVID Study

WASE-COVID Investigators

doi:10.1016/j.echo.2022.07.004

Human versus Artificial Intelligence–Based Echocardiographic Analysis as a Predictor of Outcomes: An Analysis from the World Alliance Societies of Echocardiography COVID Study

WASE-COVID Investigators

Medicine, Cardiology Division

Research output: Contribution to journal › Article › peer-review

19 Scopus citations

Abstract

Background: Transthoracic echocardiography is the leading cardiac imaging modality for patients admitted with COVID-19, a condition of high short-term mortality. The aim of this study was 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. Methods: Patients admitted to 13 hospitals for acute COVID-19 who underwent transthoracic echocardiography were included. Left ventricular ejection fraction (LVEF) and left ventricular longitudinal strain (LVLS) were obtained manually by multiple expert readers and by automated AI software. The ability of the manual and AI analyses to predict all-cause mortality was compared. Results: In total, 870 patients were enrolled. The mortality rate was 27.4% after a mean follow-up period of 230 ± 115 days. AI analysis had lower variability than manual analysis for both LVEF (P = .003) and LVLS (P = .005). AI-derived LVEF and LVLS were predictors of mortality in univariable and multivariable regression analysis (odds ratio, 0.974 [95% CI, 0.956-0.991; P = .003] for LVEF; odds ratio, 1.060 [95% CI, 1.019-1.105; P = .004] for LVLS), but LVEF and LVLS obtained by manual analysis were not. Direct comparison of the predictive value of AI versus manual measurements of LVEF and LVLS showed that AI was significantly better (P = .005 and P = .003, respectively). In addition, AI-derived LVEF and LVLS had more significant and stronger correlations to other objective biomarkers of acute disease than manual reads. Conclusions: AI-based analysis of LVEF and LVLS had similar feasibility as manual analysis, minimized variability, and consequently increased the statistical power to predict mortality. AI-based, but not manual, analyses were a significant predictor of in-hospital and follow-up mortality.

Original language	English (US)
Pages (from-to)	1226-1237.e7
Journal	Journal of the American Society of Echocardiography
Volume	35
Issue number	12
DOIs	https://doi.org/10.1016/j.echo.2022.07.004
State	Published - Dec 2022

Funding

This work was supported by the American Society of Echocardiography Foundation, the University of Chicago, and MedStar Health, with in-kind support from Ultromics.T.D., R.U., and G.M.W. are full-time employees of Ultromics. R.S. has received consulting fees from Ultromics and Weatherden. M.J.M. is an advisory board and speaker's bureau member for Bracco and Philips. R.M.L. is an advisory board and speaker's bureau member for Philips; and is an advisory board for Caption Health. F.M.A. has received institutional (MedStar Health) research grants from TomTec, Ultromics, GE, and Caption Health; and is an unpaid scientific advisory board member for Ultromics. This work was supported by the American Society of Echocardiography Foundation , the University of Chicago , and MedStar Health , with in-kind support from Ultromics .

Keywords

Artificial intelligence
COVID-19
Echocardiography
Left ventricular function
Machine learning
Outcomes prediction
WASE

ASJC Scopus subject areas

Radiology Nuclear Medicine and imaging
Cardiology and Cardiovascular Medicine

UN SDGs

This output contributes to the following UN Sustainable Development Goals (SDGs)

Access to Document

10.1016/j.echo.2022.07.004

Cite this

@article{1bdf3703f1d645efbb8ffb9ff59736d2,

title = "Human versus Artificial Intelligence–Based Echocardiographic Analysis as a Predictor of Outcomes: An Analysis from the World Alliance Societies of Echocardiography COVID Study",

abstract = "Background: Transthoracic echocardiography is the leading cardiac imaging modality for patients admitted with COVID-19, a condition of high short-term mortality. The aim of this study was 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. Methods: Patients admitted to 13 hospitals for acute COVID-19 who underwent transthoracic echocardiography were included. Left ventricular ejection fraction (LVEF) and left ventricular longitudinal strain (LVLS) were obtained manually by multiple expert readers and by automated AI software. The ability of the manual and AI analyses to predict all-cause mortality was compared. Results: In total, 870 patients were enrolled. The mortality rate was 27.4% after a mean follow-up period of 230 ± 115 days. AI analysis had lower variability than manual analysis for both LVEF (P = .003) and LVLS (P = .005). AI-derived LVEF and LVLS were predictors of mortality in univariable and multivariable regression analysis (odds ratio, 0.974 [95% CI, 0.956-0.991; P = .003] for LVEF; odds ratio, 1.060 [95% CI, 1.019-1.105; P = .004] for LVLS), but LVEF and LVLS obtained by manual analysis were not. Direct comparison of the predictive value of AI versus manual measurements of LVEF and LVLS showed that AI was significantly better (P = .005 and P = .003, respectively). In addition, AI-derived LVEF and LVLS had more significant and stronger correlations to other objective biomarkers of acute disease than manual reads. Conclusions: AI-based analysis of LVEF and LVLS had similar feasibility as manual analysis, minimized variability, and consequently increased the statistical power to predict mortality. AI-based, but not manual, analyses were a significant predictor of in-hospital and follow-up mortality.",

keywords = "Artificial intelligence, COVID-19, Echocardiography, Left ventricular function, Machine learning, Outcomes prediction, WASE",

author = "{WASE-COVID Investigators} and Asch, {Federico M.} and Tine Descamps and Rizwan Sarwar and Ilya Karagodin and Singulane, {Cristiane Carvalho} and Mingxing Xie and Tucay, {Edwin S.} and {Tude Rodrigues}, {Ana C.} and Vasquez-Ortiz, {Zuilma Y.} and Monaghan, {Mark J.} and {Ordonez Salazar}, {Bayardo A.} and Laurie Soulat-Dufour and Azin Alizadehasl and Atoosa Mostafavi and Antonella Moreo and Rodolfo Citro and Akhil Narang and Chun Wu and Karima Addetia and Ross Upton and Woodward, {Gary M.} and Lang, {Roberto M.} and Munoz, {Vince Ryan V.} and {De Marchi}, {Rafael Porto} and Alday-Ramirez, {Sergio M.} and Consuelo Orihuela and Anita Sadeghpour and Jonathan Breeze and Amy Hoare and Rosales, {Carlos Ixcanparij} and Ariel Cohen and Martina Milani and Ilaria Trolese and Oriana Belli and {De Chiara}, Benedetta and Michele Bellino and Giuseppe Iuliano and Yun Yang",

note = "Publisher Copyright: {\textcopyright} 2022 American Society of Echocardiography",

year = "2022",

month = dec,

doi = "10.1016/j.echo.2022.07.004",

language = "English (US)",

volume = "35",

pages = "1226--1237.e7",

journal = "Journal of the American Society of Echocardiography",

issn = "0894-7317",

publisher = "Elsevier Inc.",

number = "12",

}

Human versus Artificial Intelligence–Based Echocardiographic Analysis as a Predictor of Outcomes: An Analysis from the World Alliance Societies of Echocardiography COVID Study. / WASE-COVID Investigators.
In: Journal of the American Society of Echocardiography, Vol. 35, No. 12, 12.2022, p. 1226-1237.e7.

Research output: Contribution to journal › Article › peer-review

TY - JOUR

T1 - Human versus Artificial Intelligence–Based Echocardiographic Analysis as a Predictor of Outcomes

T2 - An Analysis from the World Alliance Societies of Echocardiography COVID Study

AU - WASE-COVID Investigators

AU - Asch, Federico M.

AU - Descamps, Tine

AU - Sarwar, Rizwan

AU - Karagodin, Ilya

AU - Singulane, Cristiane Carvalho

AU - Xie, Mingxing

AU - Tucay, Edwin S.

AU - Tude Rodrigues, Ana C.

AU - Vasquez-Ortiz, Zuilma Y.

AU - Monaghan, Mark J.

AU - Ordonez Salazar, Bayardo A.

AU - Soulat-Dufour, Laurie

AU - Alizadehasl, Azin

AU - Mostafavi, Atoosa

AU - Moreo, Antonella

AU - Citro, Rodolfo

AU - Narang, Akhil

AU - Wu, Chun

AU - Addetia, Karima

AU - Upton, Ross

AU - Woodward, Gary M.

AU - Lang, Roberto M.

AU - Munoz, Vince Ryan V.

AU - De Marchi, Rafael Porto

AU - Alday-Ramirez, Sergio M.

AU - Orihuela, Consuelo

AU - Sadeghpour, Anita

AU - Breeze, Jonathan

AU - Hoare, Amy

AU - Rosales, Carlos Ixcanparij

AU - Cohen, Ariel

AU - Milani, Martina

AU - Trolese, Ilaria

AU - Belli, Oriana

AU - De Chiara, Benedetta

AU - Bellino, Michele

AU - Iuliano, Giuseppe

AU - Yang, Yun

PY - 2022/12

Y1 - 2022/12

N2 - Background: Transthoracic echocardiography is the leading cardiac imaging modality for patients admitted with COVID-19, a condition of high short-term mortality. The aim of this study was 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. Methods: Patients admitted to 13 hospitals for acute COVID-19 who underwent transthoracic echocardiography were included. Left ventricular ejection fraction (LVEF) and left ventricular longitudinal strain (LVLS) were obtained manually by multiple expert readers and by automated AI software. The ability of the manual and AI analyses to predict all-cause mortality was compared. Results: In total, 870 patients were enrolled. The mortality rate was 27.4% after a mean follow-up period of 230 ± 115 days. AI analysis had lower variability than manual analysis for both LVEF (P = .003) and LVLS (P = .005). AI-derived LVEF and LVLS were predictors of mortality in univariable and multivariable regression analysis (odds ratio, 0.974 [95% CI, 0.956-0.991; P = .003] for LVEF; odds ratio, 1.060 [95% CI, 1.019-1.105; P = .004] for LVLS), but LVEF and LVLS obtained by manual analysis were not. Direct comparison of the predictive value of AI versus manual measurements of LVEF and LVLS showed that AI was significantly better (P = .005 and P = .003, respectively). In addition, AI-derived LVEF and LVLS had more significant and stronger correlations to other objective biomarkers of acute disease than manual reads. Conclusions: AI-based analysis of LVEF and LVLS had similar feasibility as manual analysis, minimized variability, and consequently increased the statistical power to predict mortality. AI-based, but not manual, analyses were a significant predictor of in-hospital and follow-up mortality.

AB - Background: Transthoracic echocardiography is the leading cardiac imaging modality for patients admitted with COVID-19, a condition of high short-term mortality. The aim of this study was 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. Methods: Patients admitted to 13 hospitals for acute COVID-19 who underwent transthoracic echocardiography were included. Left ventricular ejection fraction (LVEF) and left ventricular longitudinal strain (LVLS) were obtained manually by multiple expert readers and by automated AI software. The ability of the manual and AI analyses to predict all-cause mortality was compared. Results: In total, 870 patients were enrolled. The mortality rate was 27.4% after a mean follow-up period of 230 ± 115 days. AI analysis had lower variability than manual analysis for both LVEF (P = .003) and LVLS (P = .005). AI-derived LVEF and LVLS were predictors of mortality in univariable and multivariable regression analysis (odds ratio, 0.974 [95% CI, 0.956-0.991; P = .003] for LVEF; odds ratio, 1.060 [95% CI, 1.019-1.105; P = .004] for LVLS), but LVEF and LVLS obtained by manual analysis were not. Direct comparison of the predictive value of AI versus manual measurements of LVEF and LVLS showed that AI was significantly better (P = .005 and P = .003, respectively). In addition, AI-derived LVEF and LVLS had more significant and stronger correlations to other objective biomarkers of acute disease than manual reads. Conclusions: AI-based analysis of LVEF and LVLS had similar feasibility as manual analysis, minimized variability, and consequently increased the statistical power to predict mortality. AI-based, but not manual, analyses were a significant predictor of in-hospital and follow-up mortality.

KW - Artificial intelligence

KW - COVID-19

KW - Echocardiography

KW - Left ventricular function

KW - Machine learning

KW - Outcomes prediction

KW - WASE

UR - http://www.scopus.com/inward/record.url?scp=85136319184&partnerID=8YFLogxK

UR - http://www.scopus.com/inward/citedby.url?scp=85136319184&partnerID=8YFLogxK

U2 - 10.1016/j.echo.2022.07.004

DO - 10.1016/j.echo.2022.07.004

M3 - Article

C2 - 35863542

AN - SCOPUS:85136319184

SN - 0894-7317

VL - 35

SP - 1226-1237.e7

JO - Journal of the American Society of Echocardiography

JF - Journal of the American Society of Echocardiography

IS - 12

ER -

Human versus Artificial Intelligence–Based Echocardiographic Analysis as a Predictor of Outcomes: An Analysis from the World Alliance Societies of Echocardiography COVID Study

Abstract

Funding

Keywords

ASJC Scopus subject areas

UN SDGs

Access to Document

Other files and links

Fingerprint

Cite this