TY - JOUR
T1 - Rapid Exclusion of COVID Infection With the Artificial Intelligence Electrocardiogram
AU - Discover Consortium (Digital and Noninvasive Screening for COVID-19 with AI ECG Repository)
AU - Attia, Zachi I.
AU - Kapa, Suraj
AU - Dugan, Jennifer
AU - Pereira, Naveen
AU - Noseworthy, Peter A.
AU - Jimenez, Francisco Lopez
AU - Cruz, Jessica
AU - Carter, Rickey E.
AU - DeSimone, Daniel C.
AU - Signorino, John
AU - Halamka, John
AU - Chennaiah Gari, Nikhita R.
AU - Madathala, Raja Sekhar
AU - Platonov, Pyotr G.
AU - Gul, Fahad
AU - Janssens, Stefan P.
AU - Narayan, Sanjiv
AU - Upadhyay, Gaurav A.
AU - Alenghat, Francis J.
AU - Lahiri, Marc K.
AU - Dujardin, Karl
AU - Hermel, Melody
AU - Dominic, Paari
AU - Turk-Adawi, Karam
AU - Asaad, Nidal
AU - Svensson, Anneli
AU - Fernandez-Aviles, Francisco
AU - Esakof, Darryl D.
AU - Bartunek, Jozef
AU - Noheria, Amit
AU - Sridhar, Arun R.
AU - Lanza, Gaetano A.
AU - Cohoon, Kevin
AU - Padmanabhan, Deepak
AU - Pardo Gutierrez, Jose Alberto
AU - Sinagra, Gianfranco
AU - Merlo, Marco
AU - Zagari, Domenico
AU - Rodriguez Escenaro, Brenda D.
AU - Pahlajani, Dev B.
AU - Loncar, Goran
AU - Vukomanovic, Vladan
AU - Jensen, Henrik K.
AU - Farkouh, Michael E.
AU - Luescher, Thomas F.
AU - Su Ping, Carolyn Lam
AU - Peters, Nicholas S.
AU - Friedman, Paul A.
N1 - Publisher Copyright:
© 2021
PY - 2021/8
Y1 - 2021/8
N2 - Objective: To rapidly exclude severe acute respiratory syndrome coronavirus 2 (SARS-CoV-2) infection using artificial intelligence applied to the electrocardiogram (ECG). Methods: A global, volunteer consortium from 4 continents identified patients with ECGs obtained around the time of polymerase chain reaction–confirmed COVID-19 diagnosis and age- and sex-matched controls from the same sites. Clinical characteristics, polymerase chain reaction results, and raw electrocardiographic data were collected. A convolutional neural network was trained using 26,153 ECGs (33.2% COVID positive), validated with 3826 ECGs (33.3% positive), and tested on 7870 ECGs not included in other sets (32.7% positive). Performance under different prevalence values was tested by adding control ECGs from a single high-volume site. Results: The area under the curve for detection of acute COVID-19 infection in the test group was 0.767 (95% CI, 0.756 to 0.778; sensitivity, 98%; specificity, 10%; positive predictive value, 37%; negative predictive value, 91%). To more accurately reflect a real-world population, 50,905 normal controls were added to adjust the COVID prevalence to approximately 5% (2657/58,555), resulting in an area under the curve of 0.780 (95% CI, 0.771 to 0.790) with a specificity of 12.1% and a negative predictive value of 99.2%. Conclusion: Infection with SARS-CoV-2 results in electrocardiographic changes that permit the artificial intelligence–enhanced ECG to be used as a rapid screening test with a high negative predictive value (99.2%). This may permit the development of electrocardiography-based tools to rapidly screen individuals for pandemic control.
AB - Objective: To rapidly exclude severe acute respiratory syndrome coronavirus 2 (SARS-CoV-2) infection using artificial intelligence applied to the electrocardiogram (ECG). Methods: A global, volunteer consortium from 4 continents identified patients with ECGs obtained around the time of polymerase chain reaction–confirmed COVID-19 diagnosis and age- and sex-matched controls from the same sites. Clinical characteristics, polymerase chain reaction results, and raw electrocardiographic data were collected. A convolutional neural network was trained using 26,153 ECGs (33.2% COVID positive), validated with 3826 ECGs (33.3% positive), and tested on 7870 ECGs not included in other sets (32.7% positive). Performance under different prevalence values was tested by adding control ECGs from a single high-volume site. Results: The area under the curve for detection of acute COVID-19 infection in the test group was 0.767 (95% CI, 0.756 to 0.778; sensitivity, 98%; specificity, 10%; positive predictive value, 37%; negative predictive value, 91%). To more accurately reflect a real-world population, 50,905 normal controls were added to adjust the COVID prevalence to approximately 5% (2657/58,555), resulting in an area under the curve of 0.780 (95% CI, 0.771 to 0.790) with a specificity of 12.1% and a negative predictive value of 99.2%. Conclusion: Infection with SARS-CoV-2 results in electrocardiographic changes that permit the artificial intelligence–enhanced ECG to be used as a rapid screening test with a high negative predictive value (99.2%). This may permit the development of electrocardiography-based tools to rapidly screen individuals for pandemic control.
UR - http://www.scopus.com/inward/record.url?scp=85111611858&partnerID=8YFLogxK
UR - http://www.scopus.com/inward/citedby.url?scp=85111611858&partnerID=8YFLogxK
U2 - 10.1016/j.mayocp.2021.05.027
DO - 10.1016/j.mayocp.2021.05.027
M3 - Article
C2 - 34353468
AN - SCOPUS:85111611858
SN - 0025-6196
VL - 96
SP - 2081
EP - 2094
JO - Mayo Clinic proceedings
JF - Mayo Clinic proceedings
IS - 8
ER -