Cargando…

Limitations of Deep Learning Attention Mechanisms in Clinical Research: Empirical Case Study Based on the Korean Diabetic Disease Setting

BACKGROUND: Despite excellent prediction performance, noninterpretability has undermined the value of applying deep-learning algorithms in clinical practice. To overcome this limitation, attention mechanism has been introduced to clinical research as an explanatory modeling method. However, potentia...

Descripción completa

Detalles Bibliográficos
Autores principales: Kim, Junetae, Lee, Sangwon, Hwang, Eugene, Ryu, Kwang Sun, Jeong, Hanseok, Lee, Jae Wook, Hwangbo, Yul, Choi, Kui Son, Cha, Hyo Soung
Formato: Online Artículo Texto
Lenguaje:English
Publicado: JMIR Publications 2020
Materias:
Acceso en línea:https://www.ncbi.nlm.nih.gov/pmc/articles/PMC7773508/
https://www.ncbi.nlm.nih.gov/pubmed/33325832
http://dx.doi.org/10.2196/18418
_version_ 1783630056430501888
author Kim, Junetae
Lee, Sangwon
Hwang, Eugene
Ryu, Kwang Sun
Jeong, Hanseok
Lee, Jae Wook
Hwangbo, Yul
Choi, Kui Son
Cha, Hyo Soung
author_facet Kim, Junetae
Lee, Sangwon
Hwang, Eugene
Ryu, Kwang Sun
Jeong, Hanseok
Lee, Jae Wook
Hwangbo, Yul
Choi, Kui Son
Cha, Hyo Soung
author_sort Kim, Junetae
collection PubMed
description BACKGROUND: Despite excellent prediction performance, noninterpretability has undermined the value of applying deep-learning algorithms in clinical practice. To overcome this limitation, attention mechanism has been introduced to clinical research as an explanatory modeling method. However, potential limitations of using this attractive method have not been clarified to clinical researchers. Furthermore, there has been a lack of introductory information explaining attention mechanisms to clinical researchers. OBJECTIVE: The aim of this study was to introduce the basic concepts and design approaches of attention mechanisms. In addition, we aimed to empirically assess the potential limitations of current attention mechanisms in terms of prediction and interpretability performance. METHODS: First, the basic concepts and several key considerations regarding attention mechanisms were identified. Second, four approaches to attention mechanisms were suggested according to a two-dimensional framework based on the degrees of freedom and uncertainty awareness. Third, the prediction performance, probability reliability, concentration of variable importance, consistency of attention results, and generalizability of attention results to conventional statistics were assessed in the diabetic classification modeling setting. Fourth, the potential limitations of attention mechanisms were considered. RESULTS: Prediction performance was very high for all models. Probability reliability was high in models with uncertainty awareness. Variable importance was concentrated in several variables when uncertainty awareness was not considered. The consistency of attention results was high when uncertainty awareness was considered. The generalizability of attention results to conventional statistics was poor regardless of the modeling approach. CONCLUSIONS: The attention mechanism is an attractive technique with potential to be very promising in the future. However, it may not yet be desirable to rely on this method to assess variable importance in clinical settings. Therefore, along with theoretical studies enhancing attention mechanisms, more empirical studies investigating potential limitations should be encouraged.
format Online
Article
Text
id pubmed-7773508
institution National Center for Biotechnology Information
language English
publishDate 2020
publisher JMIR Publications
record_format MEDLINE/PubMed
spelling pubmed-77735082021-01-07 Limitations of Deep Learning Attention Mechanisms in Clinical Research: Empirical Case Study Based on the Korean Diabetic Disease Setting Kim, Junetae Lee, Sangwon Hwang, Eugene Ryu, Kwang Sun Jeong, Hanseok Lee, Jae Wook Hwangbo, Yul Choi, Kui Son Cha, Hyo Soung J Med Internet Res Original Paper BACKGROUND: Despite excellent prediction performance, noninterpretability has undermined the value of applying deep-learning algorithms in clinical practice. To overcome this limitation, attention mechanism has been introduced to clinical research as an explanatory modeling method. However, potential limitations of using this attractive method have not been clarified to clinical researchers. Furthermore, there has been a lack of introductory information explaining attention mechanisms to clinical researchers. OBJECTIVE: The aim of this study was to introduce the basic concepts and design approaches of attention mechanisms. In addition, we aimed to empirically assess the potential limitations of current attention mechanisms in terms of prediction and interpretability performance. METHODS: First, the basic concepts and several key considerations regarding attention mechanisms were identified. Second, four approaches to attention mechanisms were suggested according to a two-dimensional framework based on the degrees of freedom and uncertainty awareness. Third, the prediction performance, probability reliability, concentration of variable importance, consistency of attention results, and generalizability of attention results to conventional statistics were assessed in the diabetic classification modeling setting. Fourth, the potential limitations of attention mechanisms were considered. RESULTS: Prediction performance was very high for all models. Probability reliability was high in models with uncertainty awareness. Variable importance was concentrated in several variables when uncertainty awareness was not considered. The consistency of attention results was high when uncertainty awareness was considered. The generalizability of attention results to conventional statistics was poor regardless of the modeling approach. CONCLUSIONS: The attention mechanism is an attractive technique with potential to be very promising in the future. However, it may not yet be desirable to rely on this method to assess variable importance in clinical settings. Therefore, along with theoretical studies enhancing attention mechanisms, more empirical studies investigating potential limitations should be encouraged. JMIR Publications 2020-12-16 /pmc/articles/PMC7773508/ /pubmed/33325832 http://dx.doi.org/10.2196/18418 Text en ©Junetae Kim, Sangwon Lee, Eugene Hwang, Kwang Sun Ryu, Hanseok Jeong, Jae Wook Lee, Yul Hwangbo, Kui Son Choi, Hyo Soung Cha. Originally published in the Journal of Medical Internet Research (http://www.jmir.org), 16.12.2020. https://creativecommons.org/licenses/by/4.0/ This is an open-access article distributed under the terms of the Creative Commons Attribution License (https://creativecommons.org/licenses/by/4.0/), which permits unrestricted use, distribution, and reproduction in any medium, provided the original work, first published in the Journal of Medical Internet Research, is properly cited. The complete bibliographic information, a link to the original publication on http://www.jmir.org/, as well as this copyright and license information must be included.
spellingShingle Original Paper
Kim, Junetae
Lee, Sangwon
Hwang, Eugene
Ryu, Kwang Sun
Jeong, Hanseok
Lee, Jae Wook
Hwangbo, Yul
Choi, Kui Son
Cha, Hyo Soung
Limitations of Deep Learning Attention Mechanisms in Clinical Research: Empirical Case Study Based on the Korean Diabetic Disease Setting
title Limitations of Deep Learning Attention Mechanisms in Clinical Research: Empirical Case Study Based on the Korean Diabetic Disease Setting
title_full Limitations of Deep Learning Attention Mechanisms in Clinical Research: Empirical Case Study Based on the Korean Diabetic Disease Setting
title_fullStr Limitations of Deep Learning Attention Mechanisms in Clinical Research: Empirical Case Study Based on the Korean Diabetic Disease Setting
title_full_unstemmed Limitations of Deep Learning Attention Mechanisms in Clinical Research: Empirical Case Study Based on the Korean Diabetic Disease Setting
title_short Limitations of Deep Learning Attention Mechanisms in Clinical Research: Empirical Case Study Based on the Korean Diabetic Disease Setting
title_sort limitations of deep learning attention mechanisms in clinical research: empirical case study based on the korean diabetic disease setting
topic Original Paper
url https://www.ncbi.nlm.nih.gov/pmc/articles/PMC7773508/
https://www.ncbi.nlm.nih.gov/pubmed/33325832
http://dx.doi.org/10.2196/18418
work_keys_str_mv AT kimjunetae limitationsofdeeplearningattentionmechanismsinclinicalresearchempiricalcasestudybasedonthekoreandiabeticdiseasesetting
AT leesangwon limitationsofdeeplearningattentionmechanismsinclinicalresearchempiricalcasestudybasedonthekoreandiabeticdiseasesetting
AT hwangeugene limitationsofdeeplearningattentionmechanismsinclinicalresearchempiricalcasestudybasedonthekoreandiabeticdiseasesetting
AT ryukwangsun limitationsofdeeplearningattentionmechanismsinclinicalresearchempiricalcasestudybasedonthekoreandiabeticdiseasesetting
AT jeonghanseok limitationsofdeeplearningattentionmechanismsinclinicalresearchempiricalcasestudybasedonthekoreandiabeticdiseasesetting
AT leejaewook limitationsofdeeplearningattentionmechanismsinclinicalresearchempiricalcasestudybasedonthekoreandiabeticdiseasesetting
AT hwangboyul limitationsofdeeplearningattentionmechanismsinclinicalresearchempiricalcasestudybasedonthekoreandiabeticdiseasesetting
AT choikuison limitationsofdeeplearningattentionmechanismsinclinicalresearchempiricalcasestudybasedonthekoreandiabeticdiseasesetting
AT chahyosoung limitationsofdeeplearningattentionmechanismsinclinicalresearchempiricalcasestudybasedonthekoreandiabeticdiseasesetting