NAT COMMUN: Tsinghua University and Huazhong University of Science and Technology jointly developed the new crown AI diagnostic tool.

Early detection of COVID-19 based on chest CT can treat patients in a timely manner and help control the spread of the disease.
recently, researchers from Tsinghua University and Hua zhong University of Science and Technology collaborated to propose an artificial intelligence (AI) system for rapid detection of COVID-19, and carried out extensive statistical analysis of COVID-19 CT based on AI systems.
researchers developed and evaluated the system on a large data set of more than 10,000 CT for COVID-19, influenza A/B, non-viral community-acquired pneumonia (CAP) and non-pneumonia subjects.
In such a difficult multi-category diagnostic task, the researchers developed 97.81 percent AAUC for deep reel neural networks on 3,199 scans in the test queue, 92.99 percent for CC-CCII and 93.25 percent for two public data sets, MossMedData.
in a reader study involving five radiologists, the AI system outperformed all radiologists in more challenging tasks at two orders of magnitude higher than radiologists.
the diagnostic properties of chest X-rays (CXRs) were also compared with CT.
researchers also explained the depth network in detail to link the system output to CT performance.
code can be found at the following URL.
.