An automated COVID-19 triage pipeline using artificial intelligence based on chest radiographs and clinical data

Chris K. Kim; Ji Whae Choi; Zhicheng Jiao; Dongcui Wang; Jing Wu; Thomas Y. Yi; Kasey C. Halsey; Feyisope Eweje; Thi My Linh Tran; Chang Liu; Robin Wang; John Sollee; Celina Hsieh; Ken Chang; Fang Xue Yang; Ritambhara Singh; Jie Lin Ou; Raymond Y. Huang; Cai Feng; Michael D. Feldman; Tao Liu; Ji Sheng Gong; Shaolei Lu; Carsten Eickhoff; Xue Feng; Ihab Kamel; Ronnie Sebro; Michael K. Atalay; Terrance Healey; Yong Fan; Wei Hua Liao; Jianxin Wang; Harrison X. Bai

doi:10.1038/s41746-021-00546-w

An automated COVID-19 triage pipeline using artificial intelligence based on chest radiographs and clinical data

Chris K. Kim, Ji Whae Choi, Zhicheng Jiao, Dongcui Wang, Jing Wu, Thomas Y. Yi, Kasey C. Halsey, Feyisope Eweje, Thi My Linh Tran, Chang Liu, Robin Wang, John Sollee, Celina Hsieh, Ken Chang, Fang Xue Yang, Ritambhara Singh, Jie Lin Ou, Raymond Y. Huang, Cai Feng, Michael D. FeldmanTao Liu, Ji Sheng Gong, Shaolei Lu, Carsten Eickhoff, Xue Feng, Ihab Kamel, Ronnie Sebro, Michael K. Atalay, Terrance Healey, Yong Fan, Wei Hua Liao, Jianxin Wang, Harrison X. Bai

School of Medicine

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

Abstract

While COVID-19 diagnosis and prognosis artificial intelligence models exist, very few can be implemented for practical use given their high risk of bias. We aimed to develop a diagnosis model that addresses notable shortcomings of prior studies, integrating it into a fully automated triage pipeline that examines chest radiographs for the presence, severity, and progression of COVID-19 pneumonia. Scans were collected using the DICOM Image Analysis and Archive, a system that communicates with a hospital’s image repository. The authors collected over 6,500 non-public chest X-rays comprising diverse COVID-19 severities, along with radiology reports and RT-PCR data. The authors provisioned one internally held-out and two external test sets to assess model generalizability and compare performance to traditional radiologist interpretation. The pipeline was evaluated on a prospective cohort of 80 radiographs, reporting a 95% diagnostic accuracy. The study mitigates bias in AI model development and demonstrates the value of an end-to-end COVID-19 triage platform.

Original language	English (US)
Article number	5
Journal	npj Digital Medicine
Volume	5
Issue number	1
DOIs	https://doi.org/10.1038/s41746-021-00546-w
State	Published - Dec 2022

ASJC Scopus subject areas

Health Information Management
Health Informatics
Medicine (miscellaneous)
Computer Science Applications

Access to Document

10.1038/s41746-021-00546-w

Cite this

Kim, C. K., Choi, J. W., Jiao, Z., Wang, D., Wu, J., Yi, T. Y., Halsey, K. C., Eweje, F., Tran, T. M. L., Liu, C., Wang, R., Sollee, J., Hsieh, C., Chang, K., Yang, F. X., Singh, R., Ou, J. L., Huang, R. Y., Feng, C., ... Bai, H. X. (2022). An automated COVID-19 triage pipeline using artificial intelligence based on chest radiographs and clinical data. npj Digital Medicine, 5(1), Article 5. https://doi.org/10.1038/s41746-021-00546-w

Kim, CK, Choi, JW, Jiao, Z, Wang, D, Wu, J, Yi, TY, Halsey, KC, Eweje, F, Tran, TML, Liu, C, Wang, R, Sollee, J, Hsieh, C, Chang, K, Yang, FX, Singh, R, Ou, JL, Huang, RY, Feng, C, Feldman, MD, Liu, T, Gong, JS, Lu, S, Eickhoff, C, Feng, X, Kamel, I, Sebro, R, Atalay, MK, Healey, T, Fan, Y, Liao, WH, Wang, J & Bai, HX 2022, 'An automated COVID-19 triage pipeline using artificial intelligence based on chest radiographs and clinical data', npj Digital Medicine, vol. 5, no. 1, 5. https://doi.org/10.1038/s41746-021-00546-w

@article{b70d8662311340a190b38ef46fd9abf9,

title = "An automated COVID-19 triage pipeline using artificial intelligence based on chest radiographs and clinical data",

abstract = "While COVID-19 diagnosis and prognosis artificial intelligence models exist, very few can be implemented for practical use given their high risk of bias. We aimed to develop a diagnosis model that addresses notable shortcomings of prior studies, integrating it into a fully automated triage pipeline that examines chest radiographs for the presence, severity, and progression of COVID-19 pneumonia. Scans were collected using the DICOM Image Analysis and Archive, a system that communicates with a hospital{\textquoteright}s image repository. The authors collected over 6,500 non-public chest X-rays comprising diverse COVID-19 severities, along with radiology reports and RT-PCR data. The authors provisioned one internally held-out and two external test sets to assess model generalizability and compare performance to traditional radiologist interpretation. The pipeline was evaluated on a prospective cohort of 80 radiographs, reporting a 95% diagnostic accuracy. The study mitigates bias in AI model development and demonstrates the value of an end-to-end COVID-19 triage platform.",

author = "Kim, {Chris K.} and Choi, {Ji Whae} and Zhicheng Jiao and Dongcui Wang and Jing Wu and Yi, {Thomas Y.} and Halsey, {Kasey C.} and Feyisope Eweje and Tran, {Thi My Linh} and Chang Liu and Robin Wang and John Sollee and Celina Hsieh and Ken Chang and Yang, {Fang Xue} and Ritambhara Singh and Ou, {Jie Lin} and Huang, {Raymond Y.} and Cai Feng and Feldman, {Michael D.} and Tao Liu and Gong, {Ji Sheng} and Shaolei Lu and Carsten Eickhoff and Xue Feng and Ihab Kamel and Ronnie Sebro and Atalay, {Michael K.} and Terrance Healey and Yong Fan and Liao, {Wei Hua} and Jianxin Wang and Bai, {Harrison X.}",

note = "Publisher Copyright: {\textcopyright} 2022, The Author(s).",

year = "2022",

month = dec,

doi = "10.1038/s41746-021-00546-w",

language = "English (US)",

volume = "5",

journal = "npj Digital Medicine",

issn = "2398-6352",

publisher = "Nature Publishing Group",

number = "1",

}

TY - JOUR

T1 - An automated COVID-19 triage pipeline using artificial intelligence based on chest radiographs and clinical data

AU - Kim, Chris K.

AU - Choi, Ji Whae

AU - Jiao, Zhicheng

AU - Wang, Dongcui

AU - Wu, Jing

AU - Yi, Thomas Y.

AU - Halsey, Kasey C.

AU - Eweje, Feyisope

AU - Tran, Thi My Linh

AU - Liu, Chang

AU - Wang, Robin

AU - Sollee, John

AU - Hsieh, Celina

AU - Chang, Ken

AU - Yang, Fang Xue

AU - Singh, Ritambhara

AU - Ou, Jie Lin

AU - Huang, Raymond Y.

AU - Feng, Cai

AU - Feldman, Michael D.

AU - Liu, Tao

AU - Gong, Ji Sheng

AU - Lu, Shaolei

AU - Eickhoff, Carsten

AU - Feng, Xue

AU - Kamel, Ihab

AU - Sebro, Ronnie

AU - Atalay, Michael K.

AU - Healey, Terrance

AU - Fan, Yong

AU - Liao, Wei Hua

AU - Wang, Jianxin

AU - Bai, Harrison X.

PY - 2022/12

Y1 - 2022/12

N2 - While COVID-19 diagnosis and prognosis artificial intelligence models exist, very few can be implemented for practical use given their high risk of bias. We aimed to develop a diagnosis model that addresses notable shortcomings of prior studies, integrating it into a fully automated triage pipeline that examines chest radiographs for the presence, severity, and progression of COVID-19 pneumonia. Scans were collected using the DICOM Image Analysis and Archive, a system that communicates with a hospital’s image repository. The authors collected over 6,500 non-public chest X-rays comprising diverse COVID-19 severities, along with radiology reports and RT-PCR data. The authors provisioned one internally held-out and two external test sets to assess model generalizability and compare performance to traditional radiologist interpretation. The pipeline was evaluated on a prospective cohort of 80 radiographs, reporting a 95% diagnostic accuracy. The study mitigates bias in AI model development and demonstrates the value of an end-to-end COVID-19 triage platform.

AB - While COVID-19 diagnosis and prognosis artificial intelligence models exist, very few can be implemented for practical use given their high risk of bias. We aimed to develop a diagnosis model that addresses notable shortcomings of prior studies, integrating it into a fully automated triage pipeline that examines chest radiographs for the presence, severity, and progression of COVID-19 pneumonia. Scans were collected using the DICOM Image Analysis and Archive, a system that communicates with a hospital’s image repository. The authors collected over 6,500 non-public chest X-rays comprising diverse COVID-19 severities, along with radiology reports and RT-PCR data. The authors provisioned one internally held-out and two external test sets to assess model generalizability and compare performance to traditional radiologist interpretation. The pipeline was evaluated on a prospective cohort of 80 radiographs, reporting a 95% diagnostic accuracy. The study mitigates bias in AI model development and demonstrates the value of an end-to-end COVID-19 triage platform.

UR - http://www.scopus.com/inward/record.url?scp=85123109254&partnerID=8YFLogxK

UR - http://www.scopus.com/inward/citedby.url?scp=85123109254&partnerID=8YFLogxK

U2 - 10.1038/s41746-021-00546-w

DO - 10.1038/s41746-021-00546-w

M3 - Article

C2 - 35031687

AN - SCOPUS:85123109254

SN - 2398-6352

VL - 5

JO - npj Digital Medicine

JF - npj Digital Medicine

IS - 1

M1 - 5

ER -

An automated COVID-19 triage pipeline using artificial intelligence based on chest radiographs and clinical data

Abstract

ASJC Scopus subject areas

Access to Document

Other files and links

Fingerprint

Cite this