Comparative analysis of machine learning algorithms for computer-assisted reporting based on fully automated cross-lingual RadLex mappings

Abstract Computer-assisted reporting (CAR) tools were suggested to improve radiology report quality by context-sensitively recommending key imaging biomarkers. However, studies evaluating machine learning (ML) algorithms on cross-lingual ontological (RadLex) mappings for developing embedded CAR algo...

Descripción completa

Guardado en:

Detalles Bibliográficos
Autores principales:	Máté E. Maros, Chang Gyu Cho, Andreas G. Junge, Benedikt Kämpgen, Victor Saase, Fabian Siegel, Frederik Trinkmann, Thomas Ganslandt, Christoph Groden, Holger Wenz
Formato:	article
Lenguaje:	EN
Publicado:	Nature Portfolio 2021
Materias:	Medicine R Science Q
Acceso en línea:	https://doaj.org/article/12348a5318d740d1a6395bfc7358ede7
Etiquetas:	Agregar Etiqueta Sin Etiquetas, Sea el primero en etiquetar este registro!

id	oai:doaj.org-article:12348a5318d740d1a6395bfc7358ede7
record_format	dspace
spelling	oai:doaj.org-article:12348a5318d740d1a6395bfc7358ede72021-12-02T15:53:46ZComparative analysis of machine learning algorithms for computer-assisted reporting based on fully automated cross-lingual RadLex mappings10.1038/s41598-021-85016-92045-2322https://doaj.org/article/12348a5318d740d1a6395bfc7358ede72021-03-01T00:00:00Zhttps://doi.org/10.1038/s41598-021-85016-9https://doaj.org/toc/2045-2322Abstract Computer-assisted reporting (CAR) tools were suggested to improve radiology report quality by context-sensitively recommending key imaging biomarkers. However, studies evaluating machine learning (ML) algorithms on cross-lingual ontological (RadLex) mappings for developing embedded CAR algorithms are lacking. Therefore, we compared ML algorithms developed on human expert-annotated features against those developed on fully automated cross-lingual (German to English) RadLex mappings using 206 CT reports of suspected stroke. Target label was whether the Alberta Stroke Programme Early CT Score (ASPECTS) should have been provided (yes/no:154/52). We focused on probabilistic outputs of ML-algorithms including tree-based methods, elastic net, support vector machines (SVMs) and fastText (linear classifier), which were evaluated in the same 5 × fivefold nested cross-validation framework. This allowed for model stacking and classifier rankings. Performance was evaluated using calibration metrics (AUC, brier score, log loss) and -plots. Contextual ML-based assistance recommending ASPECTS was feasible. SVMs showed the highest accuracies both on human-extracted- (87%) and RadLex features (findings:82.5%; impressions:85.4%). FastText achieved the highest accuracy (89.3%) and AUC (92%) on impressions. Boosted trees fitted on findings had the best calibration profile. Our approach provides guidance for choosing ML classifiers for CAR tools in fully automated and language-agnostic fashion using bag-of-RadLex terms on limited expert-labelled training data.Máté E. MarosChang Gyu ChoAndreas G. JungeBenedikt KämpgenVictor SaaseFabian SiegelFrederik TrinkmannThomas GanslandtChristoph GrodenHolger WenzNature PortfolioarticleMedicineRScienceQENScientific Reports, Vol 11, Iss 1, Pp 1-18 (2021)
institution	DOAJ
collection	DOAJ
language	EN
topic	Medicine R Science Q
spellingShingle	Medicine R Science Q Máté E. Maros Chang Gyu Cho Andreas G. Junge Benedikt Kämpgen Victor Saase Fabian Siegel Frederik Trinkmann Thomas Ganslandt Christoph Groden Holger Wenz Comparative analysis of machine learning algorithms for computer-assisted reporting based on fully automated cross-lingual RadLex mappings
description	Abstract Computer-assisted reporting (CAR) tools were suggested to improve radiology report quality by context-sensitively recommending key imaging biomarkers. However, studies evaluating machine learning (ML) algorithms on cross-lingual ontological (RadLex) mappings for developing embedded CAR algorithms are lacking. Therefore, we compared ML algorithms developed on human expert-annotated features against those developed on fully automated cross-lingual (German to English) RadLex mappings using 206 CT reports of suspected stroke. Target label was whether the Alberta Stroke Programme Early CT Score (ASPECTS) should have been provided (yes/no:154/52). We focused on probabilistic outputs of ML-algorithms including tree-based methods, elastic net, support vector machines (SVMs) and fastText (linear classifier), which were evaluated in the same 5 × fivefold nested cross-validation framework. This allowed for model stacking and classifier rankings. Performance was evaluated using calibration metrics (AUC, brier score, log loss) and -plots. Contextual ML-based assistance recommending ASPECTS was feasible. SVMs showed the highest accuracies both on human-extracted- (87%) and RadLex features (findings:82.5%; impressions:85.4%). FastText achieved the highest accuracy (89.3%) and AUC (92%) on impressions. Boosted trees fitted on findings had the best calibration profile. Our approach provides guidance for choosing ML classifiers for CAR tools in fully automated and language-agnostic fashion using bag-of-RadLex terms on limited expert-labelled training data.
format	article
author	Máté E. Maros Chang Gyu Cho Andreas G. Junge Benedikt Kämpgen Victor Saase Fabian Siegel Frederik Trinkmann Thomas Ganslandt Christoph Groden Holger Wenz
author_facet	Máté E. Maros Chang Gyu Cho Andreas G. Junge Benedikt Kämpgen Victor Saase Fabian Siegel Frederik Trinkmann Thomas Ganslandt Christoph Groden Holger Wenz
author_sort	Máté E. Maros
title	Comparative analysis of machine learning algorithms for computer-assisted reporting based on fully automated cross-lingual RadLex mappings
title_short	Comparative analysis of machine learning algorithms for computer-assisted reporting based on fully automated cross-lingual RadLex mappings
title_full	Comparative analysis of machine learning algorithms for computer-assisted reporting based on fully automated cross-lingual RadLex mappings
title_fullStr	Comparative analysis of machine learning algorithms for computer-assisted reporting based on fully automated cross-lingual RadLex mappings
title_full_unstemmed	Comparative analysis of machine learning algorithms for computer-assisted reporting based on fully automated cross-lingual RadLex mappings
title_sort	comparative analysis of machine learning algorithms for computer-assisted reporting based on fully automated cross-lingual radlex mappings
publisher	Nature Portfolio
publishDate	2021
url	https://doaj.org/article/12348a5318d740d1a6395bfc7358ede7
work_keys_str_mv	AT mateemaros comparativeanalysisofmachinelearningalgorithmsforcomputerassistedreportingbasedonfullyautomatedcrosslingualradlexmappings AT changgyucho comparativeanalysisofmachinelearningalgorithmsforcomputerassistedreportingbasedonfullyautomatedcrosslingualradlexmappings AT andreasgjunge comparativeanalysisofmachinelearningalgorithmsforcomputerassistedreportingbasedonfullyautomatedcrosslingualradlexmappings AT benediktkampgen comparativeanalysisofmachinelearningalgorithmsforcomputerassistedreportingbasedonfullyautomatedcrosslingualradlexmappings AT victorsaase comparativeanalysisofmachinelearningalgorithmsforcomputerassistedreportingbasedonfullyautomatedcrosslingualradlexmappings AT fabiansiegel comparativeanalysisofmachinelearningalgorithmsforcomputerassistedreportingbasedonfullyautomatedcrosslingualradlexmappings AT frederiktrinkmann comparativeanalysisofmachinelearningalgorithmsforcomputerassistedreportingbasedonfullyautomatedcrosslingualradlexmappings AT thomasganslandt comparativeanalysisofmachinelearningalgorithmsforcomputerassistedreportingbasedonfullyautomatedcrosslingualradlexmappings AT christophgroden comparativeanalysisofmachinelearningalgorithmsforcomputerassistedreportingbasedonfullyautomatedcrosslingualradlexmappings AT holgerwenz comparativeanalysisofmachinelearningalgorithmsforcomputerassistedreportingbasedonfullyautomatedcrosslingualradlexmappings
_version_	1718385495358046208

Comparative analysis of machine learning algorithms for computer-assisted reporting based on fully automated cross-lingual RadLex mappings

Ejemplares similares