Domain Shifts in Machine Learning Based Covid-19 Diagnosis From Blood Tests

Many previous studies claim to have developed machine learning models that diagnose COVID-19 from blood tests. However, we hypothesize that changes in the underlying distribution of the data, so called domain shifts, affect the predictive performance and reliability and are a reason for the failure...

Descripción completa

Detalles Bibliográficos
Autores principales: Roland, Theresa, Böck, Carl, Tschoellitsch, Thomas, Maletzky, Alexander, Hochreiter, Sepp, Meier, Jens, Klambauer, Günter
Formato: Online Artículo Texto
Lenguaje:English
Publicado: Springer US 2022
Materias:
Clinical Systems
Acceso en línea:https://www.ncbi.nlm.nih.gov/pmc/articles/PMC8960704/
https://www.ncbi.nlm.nih.gov/pubmed/35348909
http://dx.doi.org/10.1007/s10916-022-01807-1

Internet

https://www.ncbi.nlm.nih.gov/pmc/articles/PMC8960704/
https://www.ncbi.nlm.nih.gov/pubmed/35348909
http://dx.doi.org/10.1007/s10916-022-01807-1

Ejemplares similares