There has been limited exploration of how to effectively integrate domain knowledge into machine learning for medical tabular data.Traditional approaches often rely on non-generalizable processes tailored to specific datasets.In contrast, recent advances in deep learning for language and tabular data are leading the way toward more generalizable and scalable methods of domain knowledge inclusion. In this paper, we first explore the need for domain knowledge in medical tabular data, categorize types of medical domain knowledge, and discuss how each can be leveraged in tabular machine learning. We then outline strategies for integrating this knowledge at various stages of the machine learning pipeline. Finally, building on recent advances in tabular deep learning, we propose future research directions to support the integration of domain knowledge.

There has been limited exploration of how to effectively integrate domain knowledge into machine learning for medical tabular data.Traditional approaches often rely on non-generalizable processes tailored to specific datasets.In contrast, recent advances in deep learning for language and tabular data are leading the way toward more generalizable and scalable methods of domain knowledge inclusion. In this paper, we first explore the need for domain knowledge in medical tabular data, categorize types of medical domain knowledge, and discuss how each can be leveraged in tabular machine learning. We then outline strategies for integrating this knowledge at various stages of the machine learning pipeline. Finally, building on recent advances in tabular deep learning, we propose future research directions to support the integration of domain knowledge.

Après avoir été développée en traitement automatique du langage, l’architecture Transformer s’est démocratisée dans de nombreux domaines de l’apprentissage automatique. Elle a permis de surpasser l’état de l’art dans de nombreuses tâches et a conduit à la création de très grands jeux de données afin d’améliorer les performances des modèles. En multimodalité vision-langage, les résultats encourageants des Transformers favorisent la collecte de données image-texte à très grande échelle. Cependant, il est difficile d’évaluer la qualité de ces nouveaux jeux de données, ainsi que leur influence sur la performance de ces modèles, car notre compréhension des Transformers vision-langage est encore limitée. Nous explorons les études du domaine pour mieux comprendre les processus de collecte des jeux de données, les caractéristiques de ces données et leurs impacts sur les performances des modèles.

Recent advances in vision-and-language modeling have seen the development of Transformer architectures that achieve remarkable performance on multimodal reasoning tasks.Yet, the exact capabilities of these black-box models are still poorly understood. While much of previous work has focused on studying their ability to learn meaning at the word-level, their ability to track syntactic dependencies between words has received less attention.We take a first step in closing this gap by creating a new multimodal task targeted at evaluating understanding of predicate-noun dependencies in a controlled setup.We evaluate a range of state-of-the-art models and find that their performance on the task varies considerably, with some models performing relatively well and others at chance level. In an effort to explain this variability, our analyses indicate that the quality (and not only sheer quantity) of pretraining data is essential. Additionally, the best performing models leverage fine-grained multimodal pretraining objectives in addition to the standard image-text matching objectives.This study highlights that targeted and controlled evaluations are a crucial step for a precise and rigorous test of the multimodal knowledge of vision-and-language models.