Year 2024: Trends towards Multi-[tasks, languages, models] and Large Adoption of LLMs

Summary

Objectives: This synopsis gives insights into scientific publications from 2024 in Natural Language Processing (NLP) for the biomedical domain. We present the process we followed to identify candidates for NLP's best papers and the three best papers of this year. We also analyze the current trends in the 2024 NLP-related publications.

Methods: We queried the Medline database and refined the outputs through automatic scoring. We then manually preselected publications to review and selected candidate papers through an adjudication process. External reviewers assessed the interest of those 14 candidates. At last, the section editors chose the best NLP papers.

Results: We collected 2,781 papers published in 2024, of which three were the best and selected as part of the NLP chapter. The best papers address manual annotation strategies in clinical data, claim verification in scientific literature, and generation of explanations for diagnoses in clinical cases.

Conclusion: Trends from 2024 correspond to: i) multi* frameworks, in which several languages, tasks or models are addressed within the same experiments, which is due to the availability of language models and datasets; ii) intensive exploitation of clinical data and emergence of newspapers as yet another source of information to be processed; iii)increasing number of publications issued from China and covering Chinese data, even if the processing of English data remains the most frequent; and iv) large adoption of large language models (LLMs) for various tasks.