The impact of inconsistent human annotations on AI driven clinical decision making
Abstract In supervised learning model development, domain experts are often used to provide the class labels (annotations). Annotation inconsistencies commonly occur when even highly experienced clinical experts annotate the same phenomenon (e.g., medical image, diagnostics, or prognostic status), d...
Bewaard in:
| Hoofdauteurs: | , , , |
|---|---|
| Formaat: | Artigo |
| Taal: | Inglês |
| Gepubliceerd in: |
Nature Portfolio
2023-02-01
|
| Reeks: | npj Digital Medicine |
| Online toegang: | https://doi.org/10.1038/s41746-023-00773-3 |
| Tags: |
Geen labels, Wees de eerste die dit record labelt!
|