Using explainable AI to diagnose genetic disorders

Keywords

Loading...
Thumbnail Image

Issue Date

2022-01-27

Language

en

Document type

Journal Title

Journal ISSN

Volume Title

Publisher

Title

ISSN

Volume

Issue

Startpage

Endpage

DOI

Abstract

Applying machine learning and artificial intelligence to the field of medicine could provide new insights and help diagnosing new patients. Even though these algorithms perform well, the algorithms do not have the transparency and accountability doctors have, which raises ethical concerns. In this study, the patterns in the symptoms of patients were analyzed and used to create an explainable rule-based machine learning algorithm, which could classify patients based on their symptoms. The algorithm’s logic is derivable for humans due to its rule-based design. The created program matches the performance of a decision tree classifier, but its performance does not match the performance of a support vector classifier, which is considered to be a black box algorithm where the transformations from the input to the output are obfuscated. The created rule-based program does not have obfuscated logic in its classification, which could provide reasoning behind a diagnosis. Such an explainable algorithm could eventually be used as a consulting tool for doctors when determining a possible diagnosis for a new patient with the reasoning for why such a diagnosis is suggested.

Description

Citation

Faculty

Faculteit der Sociale Wetenschappen