Interpretable AI in Healthcare

Kohler K, Kraus M (2024)


Publication Type: Book chapter / Article in edited volumes

Publication year: 2024

Publisher: CRC Press

Edited Volumes: Dimensions of Intelligent Analytics for Smart Digital Health Solutions

City/Town: New York

ISBN: 9781003849704

DOI: 10.1201/9781032699745-5

Abstract

This chapter provides a review of fundamental concepts for interpretable AI in healthcare, highlighting the currently relevant literature on healthcare analytics, intelligible models, the interpretability-accuracy trade-off, and hybrid approaches, such as explainable or interpretable machine learning models. The core concept of healthcare analytics is presented first, by illustrating the efficiency challenge in the health domain, a potential solution in the form of analytics methods performed on extensive medical data as well as the recent emergence of machine learning in the field. Second, it is attempted to define and quantify the still vague concept of model interpretability, before its necessity is evaluated from both a legislative and a practical viewpoint. Finally, this chapter recaps the interpretability-accuracy trade-off and categorises machine learning algorithms based on the transparency of their inner mechanics: presented are non-linear black-box models, simpler interpretable models, as well as hybrid model approaches, which in turn consist of the subtypes; explainable and additive hybrid models.

Authors with CRIS profile

How to cite

APA:

Kohler, K., & Kraus, M. (2024). Interpretable AI in Healthcare. In Nilmini Wickramasinghe, Freimut Bodendorf, Mathias Kraus (Eds.), Dimensions of Intelligent Analytics for Smart Digital Health Solutions. New York: CRC Press.

MLA:

Kohler, Kevin, and Mathias Kraus. "Interpretable AI in Healthcare." Dimensions of Intelligent Analytics for Smart Digital Health Solutions. Ed. Nilmini Wickramasinghe, Freimut Bodendorf, Mathias Kraus, New York: CRC Press, 2024.

BibTeX: Download