HIGHLIGHTS
SUMMARY
The authors outline and evaluate three different frameworks for adopting single-site, ready-made ML models for use in new, independent hospital settings, which are as follows: applying a ready-made model "as-is"; readjusting the decision threshold on the output of a ready-made model using site-specific data; and_(3) finetuning a ready-made model using site-specific data, by means of transfer learning. There have been different approaches used to address model generalization for clinical questions, including: combined-site training, where training data comes from multiple sites17-20; and_(2) federated . . .
If you want to have access to all the content you need to log in!
Thanks :)
If you don't have an account, you can create one here.