TY - GEN
T1 - The Pitfalls of Using Open Data to Develop Deep Learning Solutions for COVID-19 Detection in Chest X-Rays
AU - Harkness, Rachael
AU - Hall, Geoff
AU - Frangi, Alejandro F.
AU - Ravikumar, Nishant
AU - Zucker, Kieran
N1 - Funding Information:
This work uses data provided by patients and collected by the NHS as part of their care. This study was supported by AWS cloud computing credits awarded through the Diagnostic Development Initiative.
Publisher Copyright:
© 2022 International Medical Informatics Association (IMIA) and IOS Press.
PY - 2022/6/6
Y1 - 2022/6/6
N2 - Since the emergence of COVID-19, deep learning models have been developed to identify COVID-19 from chest X-rays. With little to no direct access to hospital data, the AI community relies heavily on public data comprising numerous data sources. Model performance results have been exceptional when training and testing on open-source data, surpassing the reported capabilities of AI in pneumonia-detection prior to the COVID-19 outbreak. In this study impactful models are trained on a widely used open-source data and tested on an external test set and a hospital dataset, for the task of classifying chest X-rays into one of three classes: COVID-19, non-COVID pneumonia and no-pneumonia. Classification performance of the models investigated is evaluated through ROC curves, confusion matrices and standard classification metrics. Explainability modules are implemented to explore the image features most important to classification. Data analysis and model evalutions show that the popular open-source dataset COVIDx is not representative of the real clinical problem and that results from testing on this are inflated. Dependence on open-source data can leave models vulnerable to bias and confounding variables, requiring careful analysis to develop clinically useful/viable AI tools for COVID-19 detection in chest X-rays.
AB - Since the emergence of COVID-19, deep learning models have been developed to identify COVID-19 from chest X-rays. With little to no direct access to hospital data, the AI community relies heavily on public data comprising numerous data sources. Model performance results have been exceptional when training and testing on open-source data, surpassing the reported capabilities of AI in pneumonia-detection prior to the COVID-19 outbreak. In this study impactful models are trained on a widely used open-source data and tested on an external test set and a hospital dataset, for the task of classifying chest X-rays into one of three classes: COVID-19, non-COVID pneumonia and no-pneumonia. Classification performance of the models investigated is evaluated through ROC curves, confusion matrices and standard classification metrics. Explainability modules are implemented to explore the image features most important to classification. Data analysis and model evalutions show that the popular open-source dataset COVIDx is not representative of the real clinical problem and that results from testing on this are inflated. Dependence on open-source data can leave models vulnerable to bias and confounding variables, requiring careful analysis to develop clinically useful/viable AI tools for COVID-19 detection in chest X-rays.
KW - Computing Methodologies
KW - Data Science
KW - Respiratory Tract Infections
UR - http://www.scopus.com/inward/record.url?scp=85131482849&partnerID=8YFLogxK
U2 - 10.3233/SHTI220164
DO - 10.3233/SHTI220164
M3 - Conference contribution
C2 - 35673103
AN - SCOPUS:85131482849
T3 - Studies in Health Technology and Informatics
SP - 679
EP - 683
BT - MEDINFO 2021
A2 - Otero, Paula
A2 - Scott, Philip
A2 - Martin, Susan Z.
A2 - Huesing, Elaine
PB - IOS Press
T2 - 18th World Congress on Medical and Health Informatics: One World, One Health - Global Partnership for Digital Innovation, MEDINFO 2021
Y2 - 2 October 2021 through 4 October 2021
ER -
