Lessons from shortcomings in machine learning for medical imaging - Archive ouverte HAL Accéder directement au contenu
Rapport Année : 2023

Lessons from shortcomings in machine learning for medical imaging

Résumé

Machine learning for medical imaging data has many opportunities for improving patients’ health, and has attracted a lot of attention in recent years. However, the progress of the field as a whole is being slowed down by the current incentives in (machine learning) research. In this report we summarize our findings based on literature and our own analysis, namely that larger datasets and more deep learning algorithms do not yet provide practical improvements in clinical problems. We provide recommendations for practices to adopt within research communities, as well as what we believe needs to change within research policy, to increase the impact of artificial intelligence in this field.
Fichier principal
Vignette du fichier
paper_oecd.pdf (294.81 Ko) Télécharger le fichier
paper_oecd (1).pdf (294.81 Ko) Télécharger le fichier
Origine : Fichiers produits par l'(les) auteur(s)

Dates et versions

hal-04342020 , version 1 (13-12-2023)

Identifiants

Citer

Gaël Varoquaux, Veronika Cheplygina. Lessons from shortcomings in machine learning for medical imaging. OECD. 2023. ⟨hal-04342020⟩
40 Consultations
23 Téléchargements

Altmetric

Partager

Gmail Facebook X LinkedIn More