BoxLM: Unifying Structures and Semantics of Medical Concepts for Diagnosis Prediction in Healthcare

Yanchao Tan, Hang Lv, Yunfei Zhan, Guofang Ma, Bo Xiong, Carl Yang
Proceedings of the 42nd International Conference on Machine Learning, PMLR 267:58603-58620, 2025.

Abstract

Language Models (LMs) have advanced diagnosis prediction by leveraging the semantic understanding of medical concepts in Electronic Health Records (EHRs). Despite these advancements, existing LM-based methods often fail to capture the structures of medical concepts (e.g., hierarchy structure from domain knowledge). In this paper, we propose BoxLM, a novel framework that unifies the structures and semantics of medical concepts for diagnosis prediction. Specifically, we propose a structure-semantic fusion mechanism via box embeddings, which integrates both ontology-driven and EHR-driven hierarchical structures with LM-based semantic embeddings, enabling interpretable medical concept representations. Furthermore, in the box-aware diagnosis prediction module, an evolve-and-memorize patient box learning mechanism is proposed to model the temporal dynamics of patient visits, and a volume-based similarity measurement is proposed to enable accurate diagnosis prediction. Extensive experiments demonstrate that BoxLM consistently outperforms state-of-the-art baselines, especially achieving strong performance in few-shot learning scenarios, showcasing its practical utility in real-world clinical settings.

Cite this Paper

BibTeX
@InProceedings{pmlr-v267-tan25e, title = {{B}ox{LM}: Unifying Structures and Semantics of Medical Concepts for Diagnosis Prediction in Healthcare}, author = {Tan, Yanchao and Lv, Hang and Zhan, Yunfei and Ma, Guofang and Xiong, Bo and Yang, Carl}, booktitle = {Proceedings of the 42nd International Conference on Machine Learning}, pages = {58603--58620}, year = {2025}, editor = {Singh, Aarti and Fazel, Maryam and Hsu, Daniel and Lacoste-Julien, Simon and Berkenkamp, Felix and Maharaj, Tegan and Wagstaff, Kiri and Zhu, Jerry}, volume = {267}, series = {Proceedings of Machine Learning Research}, month = {13--19 Jul}, publisher = {PMLR}, pdf = {https://raw.githubusercontent.com/mlresearch/v267/main/assets/tan25e/tan25e.pdf}, url = {https://proceedings.mlr.press/v267/tan25e.html}, abstract = {Language Models (LMs) have advanced diagnosis prediction by leveraging the semantic understanding of medical concepts in Electronic Health Records (EHRs). Despite these advancements, existing LM-based methods often fail to capture the structures of medical concepts (e.g., hierarchy structure from domain knowledge). In this paper, we propose BoxLM, a novel framework that unifies the structures and semantics of medical concepts for diagnosis prediction. Specifically, we propose a structure-semantic fusion mechanism via box embeddings, which integrates both ontology-driven and EHR-driven hierarchical structures with LM-based semantic embeddings, enabling interpretable medical concept representations. Furthermore, in the box-aware diagnosis prediction module, an evolve-and-memorize patient box learning mechanism is proposed to model the temporal dynamics of patient visits, and a volume-based similarity measurement is proposed to enable accurate diagnosis prediction. Extensive experiments demonstrate that BoxLM consistently outperforms state-of-the-art baselines, especially achieving strong performance in few-shot learning scenarios, showcasing its practical utility in real-world clinical settings.} }
Endnote
%0 Conference Paper %T BoxLM: Unifying Structures and Semantics of Medical Concepts for Diagnosis Prediction in Healthcare %A Yanchao Tan %A Hang Lv %A Yunfei Zhan %A Guofang Ma %A Bo Xiong %A Carl Yang %B Proceedings of the 42nd International Conference on Machine Learning %C Proceedings of Machine Learning Research %D 2025 %E Aarti Singh %E Maryam Fazel %E Daniel Hsu %E Simon Lacoste-Julien %E Felix Berkenkamp %E Tegan Maharaj %E Kiri Wagstaff %E Jerry Zhu %F pmlr-v267-tan25e %I PMLR %P 58603--58620 %U https://proceedings.mlr.press/v267/tan25e.html %V 267 %X Language Models (LMs) have advanced diagnosis prediction by leveraging the semantic understanding of medical concepts in Electronic Health Records (EHRs). Despite these advancements, existing LM-based methods often fail to capture the structures of medical concepts (e.g., hierarchy structure from domain knowledge). In this paper, we propose BoxLM, a novel framework that unifies the structures and semantics of medical concepts for diagnosis prediction. Specifically, we propose a structure-semantic fusion mechanism via box embeddings, which integrates both ontology-driven and EHR-driven hierarchical structures with LM-based semantic embeddings, enabling interpretable medical concept representations. Furthermore, in the box-aware diagnosis prediction module, an evolve-and-memorize patient box learning mechanism is proposed to model the temporal dynamics of patient visits, and a volume-based similarity measurement is proposed to enable accurate diagnosis prediction. Extensive experiments demonstrate that BoxLM consistently outperforms state-of-the-art baselines, especially achieving strong performance in few-shot learning scenarios, showcasing its practical utility in real-world clinical settings.
APA
Tan, Y., Lv, H., Zhan, Y., Ma, G., Xiong, B. & Yang, C.. (2025). BoxLM: Unifying Structures and Semantics of Medical Concepts for Diagnosis Prediction in Healthcare. Proceedings of the 42nd International Conference on Machine Learning, in Proceedings of Machine Learning Research 267:58603-58620 Available from https://proceedings.mlr.press/v267/tan25e.html.

Related Material