Interpretable Machine Learning Models for Healthcare Applications

EasyChair Preprint 12358

Abstract

This paper explores the importance, challenges, and applications of interpretable machine learning models in healthcare settings. The paper begins by highlighting the significance of interpretable ML models in healthcare, emphasizing the need for models that not only achieve high predictive performance but also provide insights into their decision-making process. It discusses the ethical, regulatory, and practical considerations associated with the deployment of ML algorithms in clinical settings, underscoring the importance of model interpretability for building trust and facilitating adoption by healthcare professionals. Furthermore, the paper examines various techniques and methodologies for enhancing the interpretability of ML models, including feature importance analysis, model visualization, rule extraction, and surrogate modeling. It discusses how these approaches enable clinicians to gain insights into the factors influencing model predictions and understand the underlying mechanisms driving decision-making.

Keyphrases: Applications, Healthcare, models

@booklet{EasyChair:12358,
  author    = {Julia Anderson and Jhon Thomas},
  title     = {Interpretable Machine Learning Models for Healthcare Applications},
  howpublished = {EasyChair Preprint 12358},
  year      = {EasyChair, 2024}}