Layer-Wise Relevance Propagation in Multi-label Neural Networks to Identify COVID-19 Associated Coinfections

Marilyn Bello*, Yaumara Aguilera, Gonzalo Nápoles, María M. García, Rafael Bello, Koen Vanhoof

*Corresponding author for this work

    Research output: Chapter in Book/Report/Conference proceedingConference contributionScientificpeer-review

    4 Citations (Scopus)

    Abstract

    COVID-19 has been affected worldwide since the end of 2019. Clinical studies have shown that a factor that increases its lethality is the existence of secondary infections. Coinfections associated with the infection SARS-CoV-2 are classified into bacterial infections and fungal infections. A patient may develop one, both, or neither. From a machine learning point of view, this is considered a multi-label classification problem. In this work, we propose a multi-label neural network able to detect such infections in a patient with SARS-CoV-2 and thus provide the medical community with a diagnosis to guide therapy in these patients. However, neural networks are often considered a “black box” model, as their strength in modeling complex interactions, also make their operation almost impossible to explain. Therefore, we propose three adaptations of the Layer-wise Relevance Propagation algorithm to explain multi-label neural networks. The inclusion of this post-hoc interpretability stage made it possible to identify significant input variables in a classifier output.

    Original languageEnglish
    Title of host publicationProgress in Artificial Intelligence and Pattern Recognition - 7th International Workshop on Artificial Intelligence and Pattern Recognition, IWAIPR 2021, Proceedings
    EditorsYanio Hernández Heredia, Vladimir Milián Núñez, José Ruiz Shulcloper
    PublisherSpringer Science and Business Media Deutschland GmbH
    Pages3-12
    Number of pages10
    ISBN (Print)9783030896904
    DOIs
    Publication statusPublished - 2021
    Event7th International Workshop on Artificial Intelligence and Pattern Recognition, IWAIPR 2021 - Virtual, Online
    Duration: 5 Oct 20217 Oct 2021

    Publication series

    NameLecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics)
    Volume13055 LNCS
    ISSN (Print)0302-9743
    ISSN (Electronic)1611-3349

    Conference

    Conference7th International Workshop on Artificial Intelligence and Pattern Recognition, IWAIPR 2021
    CityVirtual, Online
    Period5/10/217/10/21

    Keywords

    • Coinfections
    • COVID-19
    • Layer-wise Relevance Propagation
    • Multi-label scenario
    • Neural networks

    Fingerprint

    Dive into the research topics of 'Layer-Wise Relevance Propagation in Multi-label Neural Networks to Identify COVID-19 Associated Coinfections'. Together they form a unique fingerprint.

    Cite this