Recently, federated learning was used to massively optimize a machine learning model for COVID-19 diagnosis. In December 2021 a paper titled "Advancing COVID-19 diagnosis with privacy-preserving collaboration in artificial intelligence" was released in the peer-reviewed journal Nature Machine Intelligence. This paper presents a considerable improvement for classifying patients as Covid positive or not, by combining the data of multiple geographic locations around the world for the diagnosis of COVID-19. Key findings established how Federated learning massively improved values of sensitivity, specificity, and Area Under the Curve (AUC) for COVID-19 diagnosis. Mainly one of the best achievements of this publication is how Federated learning allowed institutions from China and the United Kingdom (UK) to cooperate together while keeping their data private and protected.
Jan-19-2022, 14:53:09 GMT