Cargando…
Eliminating biasing signals in lung cancer images for prognosis predictions with deep learning
Deep learning has shown remarkable results for image analysis and is expected to aid individual treatment decisions in health care. Treatment recommendations are predictions with an inherently causal interpretation. To use deep learning for these applications in the setting of observational data, de...
Autores principales: | , , , , |
---|---|
Formato: | Online Artículo Texto |
Lenguaje: | English |
Publicado: |
Nature Publishing Group UK
2019
|
Materias: | |
Acceso en línea: | https://www.ncbi.nlm.nih.gov/pmc/articles/PMC6904461/ https://www.ncbi.nlm.nih.gov/pubmed/31840093 http://dx.doi.org/10.1038/s41746-019-0194-x |
_version_ | 1783478007471538176 |
---|---|
author | van Amsterdam, W. A. C. Verhoeff, J. J. C. de Jong, P. A. Leiner, T. Eijkemans, M. J. C. |
author_facet | van Amsterdam, W. A. C. Verhoeff, J. J. C. de Jong, P. A. Leiner, T. Eijkemans, M. J. C. |
author_sort | van Amsterdam, W. A. C. |
collection | PubMed |
description | Deep learning has shown remarkable results for image analysis and is expected to aid individual treatment decisions in health care. Treatment recommendations are predictions with an inherently causal interpretation. To use deep learning for these applications in the setting of observational data, deep learning methods must be made compatible with the required causal assumptions. We present a scenario with real-world medical images (CT-scans of lung cancer) and simulated outcome data. Through the data simulation scheme, the images contain two distinct factors of variation that are associated with survival, but represent a collider (tumor size) and a prognostic factor (tumor heterogeneity), respectively. When a deep network would use all the information available in the image to predict survival, it would condition on the collider and thereby introduce bias in the estimation of the treatment effect. We show that when this collider can be quantified, unbiased individual prognosis predictions are attainable with deep learning. This is achieved by (1) setting a dual task for the network to predict both the outcome and the collider and (2) enforcing a form of linear independence of the activation distributions of the last layer. Our method provides an example of combining deep learning and structural causal models to achieve unbiased individual prognosis predictions. Extensions of machine learning methods for applications to causal questions are required to attain the long-standing goal of personalized medicine supported by artificial intelligence. |
format | Online Article Text |
id | pubmed-6904461 |
institution | National Center for Biotechnology Information |
language | English |
publishDate | 2019 |
publisher | Nature Publishing Group UK |
record_format | MEDLINE/PubMed |
spelling | pubmed-69044612019-12-13 Eliminating biasing signals in lung cancer images for prognosis predictions with deep learning van Amsterdam, W. A. C. Verhoeff, J. J. C. de Jong, P. A. Leiner, T. Eijkemans, M. J. C. NPJ Digit Med Article Deep learning has shown remarkable results for image analysis and is expected to aid individual treatment decisions in health care. Treatment recommendations are predictions with an inherently causal interpretation. To use deep learning for these applications in the setting of observational data, deep learning methods must be made compatible with the required causal assumptions. We present a scenario with real-world medical images (CT-scans of lung cancer) and simulated outcome data. Through the data simulation scheme, the images contain two distinct factors of variation that are associated with survival, but represent a collider (tumor size) and a prognostic factor (tumor heterogeneity), respectively. When a deep network would use all the information available in the image to predict survival, it would condition on the collider and thereby introduce bias in the estimation of the treatment effect. We show that when this collider can be quantified, unbiased individual prognosis predictions are attainable with deep learning. This is achieved by (1) setting a dual task for the network to predict both the outcome and the collider and (2) enforcing a form of linear independence of the activation distributions of the last layer. Our method provides an example of combining deep learning and structural causal models to achieve unbiased individual prognosis predictions. Extensions of machine learning methods for applications to causal questions are required to attain the long-standing goal of personalized medicine supported by artificial intelligence. Nature Publishing Group UK 2019-12-10 /pmc/articles/PMC6904461/ /pubmed/31840093 http://dx.doi.org/10.1038/s41746-019-0194-x Text en © The Author(s) 2019 Open Access This article is licensed under a Creative Commons Attribution 4.0 International License, which permits use, sharing, adaptation, distribution and reproduction in any medium or format, as long as you give appropriate credit to the original author(s) and the source, provide a link to the Creative Commons license, and indicate if changes were made. The images or other third party material in this article are included in the article’s Creative Commons license, unless indicated otherwise in a credit line to the material. If material is not included in the article’s Creative Commons license and your intended use is not permitted by statutory regulation or exceeds the permitted use, you will need to obtain permission directly from the copyright holder. To view a copy of this license, visit http://creativecommons.org/licenses/by/4.0/. |
spellingShingle | Article van Amsterdam, W. A. C. Verhoeff, J. J. C. de Jong, P. A. Leiner, T. Eijkemans, M. J. C. Eliminating biasing signals in lung cancer images for prognosis predictions with deep learning |
title | Eliminating biasing signals in lung cancer images for prognosis predictions with deep learning |
title_full | Eliminating biasing signals in lung cancer images for prognosis predictions with deep learning |
title_fullStr | Eliminating biasing signals in lung cancer images for prognosis predictions with deep learning |
title_full_unstemmed | Eliminating biasing signals in lung cancer images for prognosis predictions with deep learning |
title_short | Eliminating biasing signals in lung cancer images for prognosis predictions with deep learning |
title_sort | eliminating biasing signals in lung cancer images for prognosis predictions with deep learning |
topic | Article |
url | https://www.ncbi.nlm.nih.gov/pmc/articles/PMC6904461/ https://www.ncbi.nlm.nih.gov/pubmed/31840093 http://dx.doi.org/10.1038/s41746-019-0194-x |
work_keys_str_mv | AT vanamsterdamwac eliminatingbiasingsignalsinlungcancerimagesforprognosispredictionswithdeeplearning AT verhoeffjjc eliminatingbiasingsignalsinlungcancerimagesforprognosispredictionswithdeeplearning AT dejongpa eliminatingbiasingsignalsinlungcancerimagesforprognosispredictionswithdeeplearning AT leinert eliminatingbiasingsignalsinlungcancerimagesforprognosispredictionswithdeeplearning AT eijkemansmjc eliminatingbiasingsignalsinlungcancerimagesforprognosispredictionswithdeeplearning |