Elamur Research⌕
☰
home / papers / 2026 / From Regulatory Approvals to Patents: Cross-Domain Linking for Cardiovascular Device Traceability

From Regulatory Approvals to Patents: Cross-Domain Linking for Cardiovascular Device Traceability

Paper · 2026 · Qingqing Yang, Haijiang Liu and Moyan Li

Summary

The paper addresses the task of connecting FDA-approved medical devices to the USPTO patents that protect their technology. No official cross-reference exists between these two systems. The core difficulty is a vocabulary gap: regulatory summaries describe clinical use, while patents describe engineering mechanisms. Verified pairs share a Jaccard overlap of only 0.039.

The authors frame the problem as cross-domain entity linking with scarce labels and build a cardiovascular benchmark. It contains 434 Class III PMA devices, 698,191 patents, and 585 expert-verified pairs taken from litigation records, virtual patent marking and SEC filings.

Their framework, Bridge-MedDevKG, works coarse-to-fine in three steps:

  • The MedDevOnto ontology highlights device-critical anchor terms and maps concepts to UMLS in three tiers.
  • Candidates are generated from three combined signals: ties to the same company, closeness in meaning, and shared entities weighted by the ontology.
  • A reranker using cross-encoder scores and XGBoost trained on hard negatives removes noise.

The system reaches 91.6% recall, which is a conservative lower bound, and reduces noise by 50.9%, outperforming LLM baselines. It also produces MedDevKG, a graph of 6.8M links that supports recall root-cause analysis, IP discovery after acquisitions and technology tracking.

Key points

  • Benchmark statistics:
    • Devices are PMA approvals from 1976–2024.
    • Patents were selected by cardiovascular CPC codes such as A61F2 and A61M25.
    • A 29,758-entity company dictionary handles subsidiaries and M&A history.
    • Gold pairs cover only 88 devices (20.3%), with a median of 5 and a maximum of 83 patents per device.
  • Recall-noise trade-off: Stage 2 candidate generation already reaches 98.97% recall. Stage 3 reranking gives up 7.4 points of recall to cut noise by 50.9%, and cross-encoder scores account for just 4.0% of feature importance.
  • Per-category results: Recall stays between 89.7% and 92.3% for stents, catheters, valves and pacemakers/ICDs. The 18 anchor terms are calibrated for cardiology and would need recalibration for other specialties.
  • Error analysis: Of the 49 missed gold pairs, 60.9% are weak associations, 30.4% are genuine semantic failures and 8.7% have doubtful gold labels. Because the gold standard is incomplete, both recall and precision are hard to measure exactly.
  • Availability: Code is released at github.com/myqqhub/Bridge-MedDevKG. The authors caution that the links are correlational and should not serve as the sole basis for legal or investment decisions.

Our summary of the paper, in our own words — not its text. The work belongs to its authors; read it at the links above.

 Institutions

🎓The Hong Kong University of Science and Technology (Guangzhou)Guangzhou · China67 works · 1,765 citations🎓Wuhan University of Science and TechnologyWuhan · China3 works · 54 citations

 Fields, topics and concepts

📚 Shelves 6The field of science this work belongs to (OpenAlex fields).Computer Science · 18,614 worksArtificial Intelligence · 17,965 worksArts and Humanities · 17,450 worksLanguage and Linguistics · 17,296 worksBiochemistry, Genetics and Molecular Biology · 263 worksMolecular Biology · 260 works

Cite

Cite as Qingqing Yang, Haijiang Liu and Moyan Li. 2026. From Regulatory Approvals to Patents: Cross-Domain Linking for Cardiovascular Device Traceability. Proceedings of the 64th Annual Meeting of the Association for Computational Linguistics (Volume 1: Long Papers).

The paper itself, not this page. BibTeX from the library's record of the work. DOI 10.18653/v1/2026.acl-long.1611

@inproceedings{yang2026regulatory,
  title = {{From Regulatory Approvals to Patents: Cross-Domain Linking for Cardiovascular Device Traceability}},
  author = {Qingqing Yang, Haijiang Liu and Moyan Li},
  booktitle = {Proceedings of the 64th Annual Meeting of the Association for Computational Linguistics (Volume 1: Long Papers)},
  year = {2026},
  doi = {10.18653/v1/2026.acl-long.1611},
  url = {https://doi.org/10.18653/v1/2026.acl-long.1611}
}

References

29 references, 9 of them with a card of ours. ↗ leads to the work, «find» searches for one we have no id for yet.

  1. Josh Achiam, Steven Adler, Sandhini Agarwal, Lama Ahmad, Ilge Akkaya, Florencia Leoni Aleman, Diogo Almeida, Janko Altenschmidt, Sam Altman, Shyamal Anadkat, and 1 others. 2023. Gpt-4 technical report. arXiv preprint arXiv:2303.08774.
  2. Philip Blair and Kfir Bar. 2022. Improving few-shot domain transfer for named entity disambiguation with pattern exploitation. In Findings of the Association for Computational Linguistics: EMNLP 2022, pages 6797–6810. find
  3. Olivier Bodenreider. 2004. The unified medical language system (umls): integrating biomedical terminology. Nucleic acids research, 32(suppl_1):D267– D270. find
  4. Payal Chandak, Kexin Huang, and Marinka Zitnik. 2023. Building a knowledge graph to enable precision medicine. Scientific Data, 10(1):67.
  5. Colleen Cunningham and David Hall. 2025. Linking medical device technologies and product markets. Working paper, May 2025. find
  6. Vladimir Karpukhin, Barlas Oguz, Sewon Min, Patrick SH Lewis, Ledell Wu, Sergey Edunov, Danqi Chen, and Wen-tau Yih. 2020. Dense passage retrieval for open-domain question answering. In EMNLP (1), pages 6769–6781.
  7. Weize Kong, Swaraj Khadanga, Cheng Li, Shaleen Kumar Gupta, Mingyang Zhang, Wensong Xu, and Michael Bendersky. 2022. Multi-aspect dense retrieval. In Proceedings of the 28th ACM SIGKDD Conference on Knowledge Discovery and Data Mining, pages 3178–3186. find
  8. Jinhyuk Lee, Wonjin Yoon, Sungdong Kim, Donghyeon Kim, Sunkyu Kim, Chan Ho So, and Jaewoo Kang. 2020. Biobert: a pre-trained biomedical language representation model for biomedical text mining. Bioinformatics, 36(4):1234–1240.
  9. Fangyu Liu, Ehsan Shareghi, Zaiqiao Meng, Marco Basaldella, and Nigel Collier. 2021. Self-alignment pretraining for biomedical entity representations. In Proceedings of the 2021 conference of the North American chapter of the association for computational linguistics: human language technologies, pa…
  10. Weiwei Sun, Lingyong Yan, Xinyu Ma, Shuaiqiang Wang, Pengjie Ren, Zhumin Chen, Dawei Yin, and Zhaochun Ren. 2023. Is chatgpt good at search? investigating large language models as re-ranking agents. arXiv preprint arXiv:2304.09542.
  11. Rodrigo Nogueira and Kyunghyun Cho. 2019. Passage re-ranking with bert. arXiv preprint arXiv:1901.04085.
  12. Mike Uschold and Michael Gruninger. 1996. Ontologies: Principles, methods and applications. The knowledge engineering review, 11(2):93–136. find
  13. Eleni Partalidou, Despina Christou, and Grigorios Tsoumakas. 2022. Improving zero-shot entity retrieval through effective dense representations. In Proceedings of the 12th Hellenic Conference on Artificial Intelligence, pages 1–5. find
  14. Bo Wang, Jie Tang, Wei Fan, Songcan Chen, Zi Yang, and Yanzhu Liu. 2009. Heterogeneous cross domain ranking in latent space. In Proceedings of the 18th ACM conference on Information and knowledge management, pages 987–996. find
  15. Chintan O Patel and James J Cimino. 2006. Mining cross-terminology links in the umls. In AMIA Annual Symposium Proceedings, volume 2006, page 624. find
  16. Jian Xu, Chao Yu, Jiawei Xu, Vetle I Torvik, Jaewoo Kang, Mujeen Sung, Min Song, Yi Bu, and Ying Ding. 2025. Pubmed knowledge graph 2.0: Connecting papers, patents, and clinical trials in biomedical science. Scientific Data, 12(1):1018. find
  17. Zhen Qin, Rolf Jagerman, Kai Hui, Honglei Zhuang, Junru Wu, Le Yan, Jiaming Shen, Tianqi Liu, Jialu Liu, Donald Metzler, and 1 others. 2024. Large language models are effective text rankers with pairwise ranking prompting. In Findings of the Association for Computational Linguistics: NAACL 2024, pag…
  18. Natraj Raman, Sameena Shah, and Manuela Veloso. 2022. Structure and semantics preserving document representations. In Proceedings of the 45th International ACM SIGIR Conference on Research and Development in Information Retrieval, pages 780– 790. find
  19. Nils Reimers and Iryna Gurevych. 2019. Sentence-bert: Sentence embeddings using siamese bert-networks. In Proceedings of the 2019 Conference on Empirical Methods in Natural Language Processing and the 9th International Joint Conference on Natural Language Processing (EMNLP-IJCNLP), page 3982. Associ…
  20. Shruti Rijhwani, Jiateng Xie, Graham Neubig, and Jaime Carbonell. 2019. Zero-shot neural transfer for cross-lingual entity linking. In Proceedings of the AAAI Conference on Artificial Intelligence, volume 33, pages 6924–6931. find
  21. Stephen Robertson, Hugo Zaragoza, and 1 others. 2009. The probabilistic relevance framework: Bm25 and beyond. Foundations and Trends® in Information Retrieval, 3(4):333–389. find
  22. Jiyun Shi, Zhimeng Yuan, Wenxuan Guo, Chen Ma, Jiehao Chen, and Meihui Zhang. 2023. Knowledge-graph-enabled biomedical entity linking: a survey. World Wide Web, 26(5):2593–2622. find
  23. L Siddharth, Lucienne TM Blessing, Kristin L Wood, and Jianxi Luo. 2022. Engineering knowledge graph from patent database. Journal of Computing and Information Science in Engineering, 22(2):021008. find
  24. Hassan Soliman. 2022. Cross-domain neural entity linking. arXiv preprint arXiv:2210.15616. — [arXiv:2210.15616](https://arxiv.org/abs/2210.15616) find
  25. Zheng Yuan, Yijia Liu, Chuanqi Tan, Songfang Huang, and Fei Huang. 2021. Improving biomedical pre-trained language models with knowledge. arXiv preprint arXiv:2104.10344. — [arXiv:2104.10344](https://arxiv.org/abs/2104.10344) find
  26. Yifan Zeng, Ojas Tendolkar, Raymond Baartmans, Qingyun Wu, Lizhong Chen, and Huazheng Wang. 2024. Llm-rankfusion: Mitigating intrinsic inconsistency in llm-based ranking. arXiv preprint arXiv:2406.00231. — [arXiv:2406.00231](https://arxiv.org/abs/2406.00231) find
  27. Jingtao Zhan, Qingyao Ai, Yiqun Liu, Jiaxin Mao, Xiaohui Xie, Min Zhang, and Shaoping Ma. 2022. Disentangled modeling of domain and relevance for adaptable dense retrieval. arXiv preprint arXiv:2208.05753. — [arXiv:2208.05753](https://arxiv.org/abs/2208.05753) find
  28. Longhui Zhang, Yanzhao Zhang, Dingkun Long, Pengjun Xie, Meishan Zhang, and Min Zhang. 2023. Rankinggpt: Empowering large language models in text ranking with progressive enhancement. CoRR. find
  29. Haoyu Zuo, Yuan Yin, and Peter Childs. 2022. Patent-kg: Patent knowledge graph extraction for engineering design. Proceedings of the Design Society, 2:821– 830. find

Updated

↑ back to top

This page in Markdown → https://md.research.elamur.ai/free/en/papers/yang-2026-regulatory-approvals-patents-cross-domain/