@inproceedings{vassileva-etal-2026-enigma,
    title = "Enigma at {\#}{SMM}4{H}{--}{H}ea{RD} 2026: Leveraging Multilingual Pre-trained Models for Clinical Named Entity Recognition",
    author = "Vassileva, Sylvia  and
      Ilieva, Plamena  and
      Kostadinov, Teodor Svetoslavov  and
      Petkova, Monika Peteva  and
      Manchevski, Daniel  and
      Doynov, Vitosh  and
      Koychev, Ivan  and
      Boytcheva, Svetla",
    editor = "Lopez-Garcia, Guillermo  and
      Gonzalez-Hernandez, Graciela",
    booktitle = "Proceedings of the 11th Social Media Mining for Health Research and Applications ({SMM}4{H}-{H}ea{RD} 2026) Workshop and Shared Tasks",
    month = jul,
    year = "2026",
    address = "San Diego, United States",
    publisher = "Association for Computational Linguistics",
    url = "https://preview.aclanthology.org/ingest-acl-workshops/2026.smm4h-1.19/",
    pages = "113--120",
    ISBN = "979-8-89176-432-3",
    abstract = "This paper addresses the MultiClinAI challenge, subtask MultiClinNER, which focuses on clinical Named Entity Recognition (NER) across seven languages: Czech, Dutch, English, Italian, Romanian, Spanish, and Swedish. The main goal of MultiClinNER is to identify and extract clinical terms specifically related to diseases, procedures, and symptoms from discharge summaries. The paper explores a variety of state-of-the-art methods, both monolingual and multilingual, ranging from pretrained, zero-shot, domain-adapted transformers to fine-tuned transformer models, and demonstrates the benefits of ensemble modeling. Data augmentation through external resources significantly enhanced the models' ability to recognize clinical entities. Both monolingual and multilingual approaches showed complementary strengths depending on the language and entity type. The average F1 score achieved across the best models for each language and category is 0.6502."
}