Training and intrinsic evaluation of lightweight word embeddings for the clinical domain in Spanish

doi: 10.3389/frai.2022.970517.

eCollection 2022.


Item in Clipboard

Carolina Chiu et al.

Front Artif Intell.



Resources for Natural Language Processing (NLP) are less numerous for languages different from English. In the clinical domain, where these resources are vital for obtaining new knowledge about human health and diseases, creating new resources for the Spanish language is imperative. One of the most common approaches in NLP is word embeddings, which are dense vector representations of a word, considering the word’s context. This vector representation is usually the first step in various NLP tasks, such as text classification or information extraction. Therefore, in order to enrich Spanish language NLP tools, we built a Spanish clinical corpus from waiting list diagnostic suspicions, a biomedical corpus from medical journals, and term sequences sampled from the Unified Medical Language System (UMLS). These three corpora can be used to compute word embeddings models from scratch using Word2vec and fastText algorithms. Furthermore, to validate the quality of the calculated embeddings, we adapted several evaluation datasets in English, including some tests that have not been used in Spanish to the best of our knowledge. These translations were validated by two bilingual clinicians following an ad hoc validation standard for the translation. Even though contextualized word embeddings nowadays receive enormous attention, their calculation and deployment require specialized hardware and giant training corpora. Our static embeddings can be used in clinical applications with limited computational resources. The validation of the intrinsic test we present here can help groups working on static and contextualized word embeddings. We are releasing the training corpus and the embeddings within this publication.


Spanish language; intrinsic evaluation; medical informatics; natural language processing; neural networks; semantic evaluation; word embeddings.

Conflict of interest statement

The authors declare that the research was conducted in the absence of any commercial or financial relationships that could be construed as a potential conflict of interest.


Figure 1

(Left) Distribution of types of specialist-identified translation inadequacies among target language lexicalizations. (Right) Intersection of agreed inadequacy types between both specialists. Each number represents the size of the set.

Similar articles


    1. Akbik A., Bergmann T., Blythe D., Rasul K., Schweter S., Vollgraf R. (2019). “FLAIR: an easy-to-use framework for state-of-the-art NLP,” in NAACL 2019, 2019 Annual Conference of the North American Chapter of the Association for Computational Linguistics (Demonstrations) (Minneapolis, MN: ), 54–59.

    1. Akhtyamova L., Martínez P., Verspoor K., Cardiff J. (2020). Testing contextualized word embeddings to improve NER in Spanish clinical case narratives. IEEE Access 8, 164717–164726. 10.1109/ACCESS.2020.3018688


    1. Báez P., Bravo-Marquez F., Dunstan J., Rojas M., Villena F. (2022). Automatic extraction of nested entities in clinical referrals in Spanish. ACM Trans. Comput. Healthcare 3, 1–22. 10.1145/3498324


    1. Báez P., Villena F., Rojas M., Durán M., Dunstan J. (2020). “The Chilean waiting list corpus: a new resource for clinical named entity recognition in Spanish,” in Proceedings of the 3rd Clinical Natural Language Processing Workshop (Association for Computational Linguistics: ) (New York, NY), 291–300. 10.18653/v1/2020.clinicalnlp-1.32


    1. Bodenreider O. (2004). The unified medical language system (UMLS): integrating biomedical terminology. Nucleic Acids Res. 32, 267–270. 10.1093/nar/gkh061




Source link

Back to top button