Translational Challenges of Biomedical Machine Learning Solutions in Clinical and Laboratory Settings

Carlos Vega*, Miroslav Kratochvil, Venkata Satagopam, Reinhard Schneider

*Corresponding author for this work

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

3 Citations (Scopus)

Abstract

The ever increasing use of artificial intelligence (AI) methods in biomedical sciences calls for closer inter-disciplinary collaborations that transfer the domain knowledge from life scientists to computer science researchers and vice-versa. We highlight two general areas where the use of AI-based solutions designed for clinical and laboratory settings has proven problematic. These are used to demonstrate common sources of translational challenges that often stem from the differences in data interpretation between the clinical and research view, and the unmatched expectations and requirements on the result quality metrics. We outline how explicit interpretable inference reporting might be used as a guide to overcome such translational challenges. We conclude with several recommendations for safer translation of machine learning solutions into real-world settings.

Original languageEnglish
Title of host publicationBioinformatics and Biomedical Engineering - 9th International Work-Conference, IWBBIO 2022, Proceedings
EditorsIgnacio Rojas, Olga Valenzuela, Fernando Rojas, Luis Javier Herrera, Francisco Ortuño
PublisherSpringer Science and Business Media Deutschland GmbH
Pages353-358
Number of pages6
ISBN (Print)9783031078019
DOIs
Publication statusPublished - 2022
Externally publishedYes
Event9th International Work-Conference on Bioinformatics and Biomedical Engineering, IWBBIO 2022 - Gran Canaria, Spain
Duration: 27 Jun 202230 Jun 2022

Publication series

NameLecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics)
Volume13347 LNBI
ISSN (Print)0302-9743
ISSN (Electronic)1611-3349

Conference

Conference9th International Work-Conference on Bioinformatics and Biomedical Engineering, IWBBIO 2022
Country/TerritorySpain
CityGran Canaria
Period27/06/2230/06/22

Keywords

  • Biomedicine
  • Machine learning

Fingerprint

Dive into the research topics of 'Translational Challenges of Biomedical Machine Learning Solutions in Clinical and Laboratory Settings'. Together they form a unique fingerprint.

Cite this