Please use this identifier to cite or link to this item:
Artificial intelligence for rapid identification of the coronavirus disease 2019 (COVID-19)
Xueyan Mei.
Hao-Chih Lee.
Kaiyue Diao.
Mingqian Huang.
Bin Lin.
Chenyu Liu.
Zongyu Xie.
Yixuan Ma.
Philip M. Robson.
Michael Chung.
Adam Bernheim.
Venkatesh Mani.
Claudia Calcagno.
Kunwei Li.
Shaolin Li.
Hong Shan.
Jian Lv.
Tongtong Zhao.
Junli Xia.
Qihua Long.
Sharon Steinberger.
Adam Jacobi.
Timothy Deyer.
Marta Luksza.
Fang Liu.
Brent P. Little.
Zahi A. Fayad.
Yang Yang.
Acceso Abierto
The coronavirus disease 2019 (COVID-19) outbreak that originated in Wuhan, China has rapidly propagated due to widespread person-to-person transmission and has resulted in over 1,133,758 cases in 197 countries with a total of 62,784 deaths as of April 5, 2020. Laboratory confirmation of SARS-CoV-2 is performed with a virus-specific reverse transcriptase polymerase chain reaction (RT-PCR) test. This test can take up to two days to complete, and, due to the possibility of false negatives, serial testing may be required to reliably exclude infection. A current supply shortage of RT-PCR test kits compounds the shortcomings of entrusting diagnosis to the PCR test alone and underscores the urgent need to provide alternative methods for the rapid and accurate diagnosis of SARS-CoV-2 patients. Chest computed tomography (CT) is a valuable component in the evaluation of patients with suspected SARS-CoV-2 infection. Nevertheless, CT alone may have limited negative predictive value to fully exclude infection, because of the normal radiologic findings in some early disease patients. In this study, we use artificial intelligence (AI) algorithms to integrate chest CT findings with clinical symptoms, exposure history, and/or laboratory testing to more accurately and rapidly diagnose SARS-CoV-2 (+) patients. We included 905 RT-PCR confirmed patients. 419 (46.2%) tested positive for SARS-CoV-2 by laboratory-confirmed real-time RT-PCR assay and next-generation sequencing, while 486 patients (53.8%) tested negative (confirmed by at least two additional negative RT-PCR tests and clinical observation). The proposed AI system achieved an AUC of 0.92 and performed equally well in sensitivity compared to a senior thoracic radiologist on a testing set of 279 cases. The AI system also improved the detection of RT-PCR positive SARS-CoV-2 patients who presented with normal CTs, correctly identifying 17/25 (68%) patients, whereas all 25 RT-PCR SARS-CoV-2-positive CT-normal patients were classified as SARS-CoV-2 negative by radiologists.
Appears in Collections:Artículos científicos

Upload archives

File SizeFormat 
1101864.pdf2.36 MBAdobe PDFView/Open