1.
Preprint
en Inglés
| medRxiv
| ID: ppmedrxiv-20035816
RESUMEN
BackgroundCOVID-19 has been spreading globally since emergence, but the diagnostic resources are relatively insufficient. ResultsIn order to effectively relieve the resource deficiency of diagnosing COVID-19, we developed a machine learning-based diagnosis model on basis of laboratory examinations indicators from a total of 620 samples, and subsequently implemented it as a COVID-19 diagnosis aid APP to facilitate promotion. ConclusionsExternal validation showed satisfiable model prediction performance (i.e., the positive predictive value and negative predictive value was 86.35% and 84.62%, respectively), which guarantees the promising use of this tool for extensive screening.