EUVIMEDEUVIMED entdecken
Open Access

Automated Large-Scale Extraction of Nanotoxicity Data from the Literature Using Multimodal Large Language Models

Seung-Geun Park iD, Seung Min Ha iD, Zayakhuu Gerelkhuu iD, Tae Hyun Yoon iD

Computational and Structural Biotechnology Journal · 01.01.2026 · 35 (3)

DOI: 10.34133/csbj.0161 ↗

Kurz gefasst

Fachartikel in der Zeitschrift Computational and Structural Biotechnology Journal, online erschienen am 07.07.2026. Verfasst von Seung-Geun Park und 3 weiteren Autor:innen. Beteiligte Einrichtung: Department of Chemistry, Hanyang University, Seoul 04763, Republic of Korea. Die Arbeit wurde bisher 3-mal zitiert (Crossref) und nennt 53 Literaturangaben. Der Volltext ist frei in PubMed Central verfügbar. Gefördert von: National Research Foundation of Korea, HORIZON EUROPE Framework Programme.

AbstractBibliografische AngabenZitierenVerwandte Studien

Abstract (Originalsprache)

Manual curation of nanotoxicity data remains a major bottleneck in the scalable construction of large datasets because key physicochemical and toxicological variables are often distributed across text, figures, tables, captions, legends, and Materials and Methods sections. Here, we developed a multimodal large language model (LLM) workflow for automated extraction and structured curation of nanotoxicity data from research articles. Four multimodal LLMs were benchmarked on curated set of 17 oxide nanotoxicity articles. GPT-5.4 provided the best balance between extraction accuracy and efficiency, achieving a mean extraction F1 score of 96.1% and an assay results F1 score of 86.5% on the main benchmark. The optimized workflow was then applied to 821 articles, generating 16,786 structured records and substantially reducing the manual recovery of figure-derived quantitative data. Automated machine learning (AutoML) modeling showed that the physicochemical (PChem) score-filtered subset achieved higher overall predictive performance, whereas record-count balancing among the 4 most represented materials reduced performance. Models trained only on records for ZnO, TiO 2 , Fe 3 O 4 , or SiO 2 improved prediction for the corresponding material compared with multi-material models, but applicability domain analysis showed that these gains occurred within narrower descriptor-space coverage. These findings demonstrate that multimodal extraction can move nanotoxicity curation beyond text-only workflows while supporting data-aware model selection across broad-coverage, quality-filtered, focused multi-material, and single-material prediction strategies.

Autor:innen und Institutionen

  • Department of Chemistry, Hanyang University, Seoul 04763, Republic of Korea.
  • Research Institute for Convergence of Basic Science, Hanyang University, Seoul 04763, Republic of Korea.
  • Institute of Next Generation Material Design, Hanyang University, Seoul 04763, Republic of Korea.
  • Yoon Idea Lab. Co., Ltd. Seoul 04763, Republic of Korea.

ORCID-Profile: Seung-Geun Park, Seung Min Ha, Zayakhuu Gerelkhuu, Tae Hyun Yoon

Förderung

  • National Research Foundation of Korea
  • HORIZON EUROPE Framework Programme

Zitieren

APA

Park, S. G., Ha, S. M., Gerelkhuu, Z., & Yoon, T. H. (2026). Automated Large-Scale Extraction of Nanotoxicity Data from the Literature Using Multimodal Large Language Models. Computational and Structural Biotechnology Journal, 35(3). https://doi.org/10.34133/csbj.0161

Vancouver

Park SG, Ha SM, Gerelkhuu Z, Yoon TH. Automated Large-Scale Extraction of Nanotoxicity Data from the Literature Using Multimodal Large Language Models. Computational and Structural Biotechnology Journal. 2026;35(3). doi:10.34133/csbj.0161

BibTeX
@article{Park2026,
  author = {Park, Seung-Geun and Ha, Seung Min and Gerelkhuu, Zayakhuu and Yoon, Tae Hyun},
  title = {Automated Large-Scale Extraction of Nanotoxicity Data from the Literature Using Multimodal Large Language Models},
  journal = {Computational and Structural Biotechnology Journal},
  year = {2026},
  volume = {35},
  number = {3},
  doi = {10.34133/csbj.0161}
}
Kostenlos starten →