Early identification of ICU patients at risk of complications: Regularization based on robustness and stability of explanations

作者：

Highlights：

• The introduction of the concepts of explanation robustness and stability, and algorithms to compute them

• A new regularization term which exploits these new concepts

• A model sampling approach which finds models with high predictive power while providing robust and stable explanations

摘要

•The introduction of the concepts of explanation robustness and stability, and algorithms to compute them•A new regularization term which exploits these new concepts•A model sampling approach which finds models with high predictive power while providing robust and stable explanations

论文关键词：Machine learning,Medical data science,Model explainability,Regularization

论文评审过程：Received 27 July 2021, Revised 14 March 2022, Accepted 17 March 2022, Available online 22 March 2022, Version of Record 2 April 2022.

论文官网地址：https://doi.org/10.1016/j.artmed.2022.102283