Model Distillation for Faithful Explanations of Medical Code Predictions

Zach Wood-Doughty, Isabel Cachola, Mark Dredze

Research output: Chapter in Book/Report/Conference proceedingConference contribution

Abstract

Machine learning models that offer excellent predictive performance often lack the interpretability necessary to support integrated human machine decision-making. In clinical or other high-risk settings, domain experts may be unwilling to trust model predictions without explanations. Work in explainable AI must balance competing objectives along two different axes: 1) Models should ideally be both accurate and simple. 2) Explanations must balance faithfulness to the model's decision-making with their plausibility to a domain expert. We propose to use knowledge distillation, or training a student model that mimics the behavior of a trained teacher model, as a technique to generate faithful and plausible explanations. We evaluate our approach on the task of assigning ICD codes to clinical notes to demonstrate that the student model is faithful to the teacher model's behavior and produces quality natural language explanations.

Original languageEnglish (US)
Title of host publicationBioNLP 2022 @ ACL 2022 - Proceedings of the 21st Workshop on Biomedical Language Processing
PublisherAssociation for Computational Linguistics (ACL)
Pages412-425
Number of pages14
ISBN (Electronic)9781955917278
StatePublished - 2022
Event21st Workshop on Biomedical Language Processing, BioNLP 2022 at the Association for Computational Linguistics Conference, ACL 2022 - Dublin, Ireland
Duration: May 26 2022 → …

Publication series

NameProceedings of the Annual Meeting of the Association for Computational Linguistics
ISSN (Print)0736-587X

Conference

Conference21st Workshop on Biomedical Language Processing, BioNLP 2022 at the Association for Computational Linguistics Conference, ACL 2022
Country/TerritoryIreland
CityDublin
Period5/26/22 → …

ASJC Scopus subject areas

  • Computer Science Applications
  • Linguistics and Language
  • Language and Linguistics

Fingerprint

Dive into the research topics of 'Model Distillation for Faithful Explanations of Medical Code Predictions'. Together they form a unique fingerprint.

Cite this