{"type":"ScholarlyArticle","title":"Structure-Aware Contrastive Learning for Biomedical Embeddings: Bridging the Gap Between HPO and Clinical Literature","authors":[{"name":"Jose Luis Mellina Andreu","slug":"jose-luis-mellina-andreu","given":"Jose Luis","family":"Mellina Andreu","affiliation":"Department of Information and Communication Engineering , University of Murcia"},{"name":"Alejandro Cisterna García","slug":"alejandro-cisterna-garcia","given":"Alejandro","family":"Cisterna García","affiliation":"Department of Health Sciences and Technology (DHEST) , ETH Zurich"},{"name":"Juan Botía","slug":"juan-botia","given":"Juan","family":"Botía","affiliation":"Department of Information and Communication Engineering , University of Murcia"}],"abstract":"Large Language Models (LLMs) are extensively used at biomedical text processing but often fail to capture the complex, functional relationships encoded in expert knowledge graphs like the Human Phenotype Ontology (HPO). This \"semantic gap'\" limits their utility in precision medicine tasks such as rare disease diagnosis, where distinguishing overlapping clinical presentations requires understanding underlying pathophysiological connections rather than just surface-level textual similarity. In this work, we propose a Neuro-Symbolic Alignment Framework that bridges this separation by integrating literature-mined specialized phenotypical descriptions with the ontological structure used as reference. Specifically, we augment phenotype representations with automatically selected text fragments from massive corpus of descriptions mined from scientific literature (PubMed), overcoming the typical data scarcity of standard ontology definitions. We define a new embedding adaptation procedure whose fine-tuning approach is guided by a novel \"Disease-Overlap\" similarity measure, which prioritizes clinical co-occurrence of phenotypes over taxonomic distance, and optimizes the embedding space using AnglE Loss to mitigate gradient saturation. Extensive evaluations show that our approach significantly outperforms state-of-the-art baselines, including SapBERT, on both intrinsic semantic correlation and practical downstream tasks, including synthetic patient disease ranking and solving real cases stored in Phenopacket, where our model achieves x4 top-1 accuracy than the previous best model.","doi":"10.24963/ijcai.2026/760","url":"https://ijcai.org/proceedings/2026/760","pdf":"https://ijcai.org/proceedings/2026/0760.pdf","markdown":"https://ijcai.org/proceedings/2026/760.md","bibtex":"https://ijcai.org/proceedings/2026/bibtex/760","volume":{"slug":"2026","title":"Proceedings of the Thirty-Fifth International Joint Conference on Artificial Intelligence","short_title":"IJCAI 2026","url":"https://ijcai.org/proceedings/2026/","year":2026,"location":"Bremen, Germany","dates":"15-21 August 2026","editors":"Diego Calvanese","isbn":"978-1-956792-09-6","publisher":"International Joint Conferences on Artificial Intelligence Organization","doi_prefix":"10.24963/ijcai.2026"},"section":"AI and Health","track":"AI and Health","number":760,"pages":{"first":6832,"last":6840},"keywords":["Biomedical NLP: Biomedical NLP","Computational phenotyping: Computational phenotyping","Medical knowledge representation: Medical knowledge representation","LLM in medicine: LLM in medicine","Clinical decision support system: Clinical decision support system"],"videos":[],"references":[{"position":1,"raw":"[Akiba and others, 2019] Takuya Akiba et al. Optuna: A next-generation hyperparameter optimization framework. In Proceedings of the 25th ACM SIGKDD International Conference on Knowledge Discovery & Data Mining, KDD '19, page 2623-2631, New York, NY, USA, 2019. Association for Computing Machinery. [Bordes and others, 2013]","title":"Optuna: A next-generation hyperparameter optimization framework","authors":"Takuya Akiba","year":"2013","venue":"Proceedings of the 25th ACM SIGKDD International Conference on Knowledge Discovery & Data Mining, KDD '19","doi":null,"url":null,"internal_url":null},{"position":2,"raw":"Antoine Bordes et al. Translating embeddings for modeling multi-relational data. In Pro- ceedings of the 27th International Conference on Neu- ral Information Processing Systems -Volume 2, NIPS'13, page 2787-2795, Red Hook, NY, USA, 2013.","title":"Translating embeddings for modeling multi-relational data","authors":"Antoine Bordes","year":"2013","venue":"Proceedings of the 27th International Conference on Neural Information Processing Systems","doi":null,"url":null,"internal_url":null},{"position":3,"raw":"Curran As- sociates Inc. [Cer and others, 2017]","title":null,"authors":"Curran Associates, Inc","year":"2017","venue":null,"doi":null,"url":null,"internal_url":null},{"position":4,"raw":"Daniel Cer et al. SemEval-2017 task 1: Semantic textual similarity multilingual and crosslin- gual focused evaluation. In Proceedings of the 11th In- ternational Workshop on Semantic Evaluation (SemEval- 2017), pages 1-14, Vancouver, Canada, August 2017. As- sociation for Computational Linguistics.","title":"SemEval-2017 task 1: Semantic textual similarity multilingual and crosslingual focused evaluation","authors":"Daniel Cer","year":"2017","venue":"Proceedings of the 11th International Workshop on Semantic Evaluation (SemEval-2017)","doi":null,"url":null,"internal_url":null},{"position":5,"raw":"[Cui et al., 2020] Baiyun Cui, Yingming Li, and Zhongfei Zhang. BERT-enhanced relational sentence ordering net- work. In Bonnie Webber, Trevor Cohn, Yulan He, and Yang Liu, editors, Proceedings of the 2020 Conference on Empirical Methods in Natural Language Processing (EMNLP), pages 6310-6320, Online, November 2020. Association for Computational Linguistics.","title":"BERT-enhanced relational sentence ordering network","authors":"Cui","year":"2020","venue":"Proceedings of the 2020 Conference on Empirical Methods in Natural Language Processing (EMNLP)","doi":null,"url":null,"internal_url":null},{"position":6,"raw":"[Danis and others, 2025] Daniel Danis et al. A corpus of GA4GH phenopackets: Case-level phenotyping for ge- nomic diagnostics and discovery. HGG Adv., 6(1):100371, January 2025.","title":"A corpus of GA4GH phenopackets: Case-level phenotyping for genomic diagnostics and discovery","authors":"Daniel Danis","year":"2025","venue":"HGG Adv","doi":null,"url":null,"internal_url":null},{"position":7,"raw":"[Enevoldsen and others, 2025] Kenneth Enevoldsen et al. Mmteb: Massive multilingual text embedding benchmark. arXiv preprint arXiv:2502.13595, 2025.","title":"Mmteb: Massive multilingual text embedding benchmark","authors":"Kenneth Enevoldsen","year":"2025","venue":null,"doi":null,"url":null,"internal_url":null},{"position":8,"raw":"[Garnot and Landrieu, 2020] Vivien Sainte Fare Garnot and Loïc Landrieu. Metric-guided prototype learning. CoRR, abs/2007.03047, 2020. Genetic Alliance and others, 2010] Genetic Alliance et al. Understanding genetics: A New England guide for pa- tients and health professionals. Genetic Alliance, Wash- ington (DC), February 2010.","title":"Vivien Sainte Fare Garnot and Loïc Landrieu. Metric-guided prototype learning","authors":"Landrieu Garnot","year":"2010","venue":"Genetic Alliance et al. Understanding genetics: A New England guide for patients and health professionals","doi":null,"url":null,"internal_url":null},{"position":9,"raw":"[Gong et al., 2018] Xiaofeng Gong, Jianping Jiang, Zhongqu Duan, and Hui Lu. A new method to mea- sure the semantic similarity from query phenotypic abnormalities to diseases based on the human phenotype ontology. BMC Bioinformatics, 19(S4), May 2018.","title":"A new method to measure the semantic similarity from query phenotypic abnormalities to diseases based on the human phenotype ontology","authors":"Xiaofeng Gong, Jianping Jiang, Zhongqu Duan, Hui Lu","year":"2018","venue":"BMC Bioinformatics","doi":null,"url":null,"internal_url":null},{"position":10,"raw":"[Grover and Leskovec, 2016] Aditya Grover and Jure Leskovec. node2vec: Scalable feature learning for networks. In Proceedings of the 22nd ACM SIGKDD International Conference on Knowledge Discovery and Data Mining, KDD '16, page 855-864, New York, NY, USA, 2016. Association for Computing Machinery.","title":"node2vec: Scalable feature learning for networks","authors":"Leskovec Grover, Aditya Grover, Jure Leskovec","year":"2016","venue":"Proceedings of the 22nd ACM SIGKDD International Conference on Knowledge Discovery and Data Mining, KDD '16","doi":null,"url":null,"internal_url":null},{"position":11,"raw":"[Gu et al., 2021] Yu Gu, Robert Tinn, Hao Cheng, Michael Lucas, Naoto Usuyama, Xiaodong Liu, Tristan Naumann, Jianfeng Gao, and Hoifung Poon. Domain-specific lan- guage model pretraining for biomedical natural language processing. ACM Trans. Comput. Healthcare, 3(1), Octo- ber 2021.","title":"Domain-specific language model pretraining for biomedical natural language processing","authors":"Yu Gu, Robert Tinn, Hao Cheng, Michael Lucas, Naoto Usuyama, Xiaodong Liu, Tristan Naumann, Jianfeng Gao, Hoifung Poon","year":"2021","venue":"ACM Trans. Comput. Healthcare","doi":null,"url":null,"internal_url":null},{"position":12,"raw":"[Hao et al., 2020] Junheng Hao, Chelsea J.-T Ju, Muhao Chen, Yizhou Sun, Carlo Zaniolo, and Wei Wang. Bio- joie: Joint representation learning of biological knowl- edge bases. In Proceedings of the 11th ACM Interna- tional Conference on Bioinformatics, Computational Bi- ology and Health Informatics, BCB '20, New York, NY, USA, 2020. Association for Computing Machinery.","title":"Biojoie: Joint representation learning of biological knowledge bases","authors":"Hao","year":"2020","venue":"Proceedings of the 11th ACM International Conference on Bioinformatics, Computational Biology and Health Informatics, BCB '20","doi":null,"url":null,"internal_url":null},{"position":13,"raw":"[Huang et al., 2024] Xiang Huang, Hao Peng, Dongcheng Zou, Zhiwei Liu, Jianxin Li, Kay Liu, Jia Wu, Jianlin Su, and Philip S. Yu. Cosent: Consistent sentence embedding via similarity ranking. IEEE/ACM Transactions on Audio, Speech, and Language Processing, 32:2800-2813, 2024.","title":"Cosent: Consistent sentence embedding via similarity ranking","authors":"Xiang Huang, Hao Peng, Dongcheng Zou, Zhiwei Liu, Jianxin Li, Kay Liu, Jia Wu, Jianlin Su, Philip S Yu","year":"2024","venue":"IEEE/ACM Transactions on Audio, Speech, and Language Processing","doi":null,"url":null,"internal_url":null},{"position":14,"raw":"[Köhler and others, 2009] Sebastian Köhler et al. Clinical diagnostics in human genetics with semantic similarity searches in ontologies. The American Journal of Human Genetics, 85(4):457-464, 2009.","title":"Clinical diagnostics in human genetics with semantic similarity searches in ontologies","authors":"Sebastian Köhler, Köhler","year":"2009","venue":"The American Journal of Human Genetics","doi":null,"url":null,"internal_url":null},{"position":15,"raw":"[Köhler and others, 2020] Sebastian Köhler et al. The hu- man phenotype ontology in 2021. Nucleic Acids Research, 49(D1):D1207-D1217, 12 2020.","title":"The human phenotype ontology in 2021","authors":"Sebastian Köhler","year":"2020","venue":"Nucleic Acids Research","doi":null,"url":null,"internal_url":null},{"position":16,"raw":"[Lee et al., 2019] Jinhyuk Lee, Wonjin Yoon, Sungdong Kim, Donghyeon Kim, Sunkyu Kim, Chan Ho So, and Jaewoo Kang. Biobert: a pre-trained biomedical language representation model for biomedical text mining. Bioin- formatics, 36(4):1234-1240, 09 2019.","title":"Biobert: a pre-trained biomedical language representation model for biomedical text mining","authors":"Lee","year":"2019","venue":"Bioinformatics","doi":null,"url":null,"internal_url":null},{"position":17,"raw":"[Li and Li, 2024] Xianming Li and Jing Li. AoE: Angle- optimized embeddings for semantic textual similarity. In Lun-Wei Ku, Andre Martins, and Vivek Srikumar, edi- tors, Proceedings of the 62nd Annual Meeting of the As- sociation for Computational Linguistics (Volume 1: Long Papers), pages 1825-1839, Bangkok, Thailand, August 2024. Association for Computational Linguistics.","title":"AoE: Angleoptimized embeddings for semantic textual similarity","authors":"Li ; Li, Xianming Li, Jing Li","year":"2024","venue":"Proceedings of the 62nd Annual Meeting of the Association for Computational Linguistics","doi":null,"url":null,"internal_url":null},{"position":18,"raw":"[Lin, 1998] Dekang Lin. An information-theoretic defini- tion of similarity. In Proceedings of the Fifteenth Interna- tional Conference on Machine Learning, ICML '98, page 296-304, San Francisco, CA, USA, 1998.","title":"An information-theoretic definition of similarity","authors":"Dekang Lin","year":"1998","venue":"Proceedings of the Fifteenth International Conference on Machine Learning, ICML '98","doi":null,"url":null,"internal_url":null},{"position":19,"raw":"Fangyu Liu et al. SapBERT: Self- alignment pretraining for biomedical entity representation. In Proceedings of the 2021 Conference of the North Ameri- can Chapter of the Association for Computational Linguis- tics: Human Language Technologies, pages 4234-4243. Association for Computational Linguistics, 2021.","title":"SapBERT: Selfalignment pretraining for biomedical entity representation","authors":"Fangyu Liu","year":"2021","venue":"Proceedings of the 2021 Conference of the North American Chapter of the Association for Computational Linguistics: Human Language Technologies","doi":null,"url":null,"internal_url":null},{"position":20,"raw":"[Lobo et al., 2017] Manuel Lobo, Andre Lamurias, and Francisco M Couto. Identifying human phenotype terms by combining machine learning and validation rules. Biomed Res. Int., 2017:1-8, 2017.","title":"Identifying human phenotype terms by combining machine learning and validation rules","authors":"Manuel Lobo, Andre Lamurias, Francisco M Couto","year":"2017","venue":"Biomed Res. Int","doi":null,"url":null,"internal_url":null},{"position":21,"raw":"[Lowe et al., 1952] Charles Lowe, Mary Terrey, and E. A. MacLachlan. Organic-aciduria, decreased renal ammo- nia production, hydrophthalmos, and mental retardation: A clinical entity. A.M.A. American Journal of Diseases of Children, 83(2):164-184, 02 1952.","title":"Organic-aciduria, decreased renal ammonia production, hydrophthalmos, and mental retardation: A clinical entity. A.M.A","authors":"Charles Lowe, Mary Terrey, E A Maclachlan","year":"1952","venue":"American Journal of Diseases of Children","doi":null,"url":null,"internal_url":null},{"position":22,"raw":"[Luo et al., 2021] Ling Luo, Shankai Yan, Po-Ting Lai, Daniel Veltri, Andrew Oler, Eyasu Goshu, Tiange Li, Anna A Morgan, Chunhua Weng, and Zhiyong Lu. Phe- notagger: a hybrid method for phenotype concept recog- nition using human phenotype ontology. Bioinformatics, 37(13):1884-1890, 2021.","title":"Phenotagger: a hybrid method for phenotype concept recognition using human phenotype ontology","authors":"Ling Luo, Shankai Yan, Po-Ting Lai, Daniel Veltri, Andrew Oler, Eyasu Goshu, Tiange Li, Anna A Morgan, Chunhua Weng, Zhiyong Lu","year":"2021","venue":"Bioinformatics","doi":null,"url":null,"internal_url":null},{"position":23,"raw":"[Marelli and others, 2014] Marco Marelli et al. A SICK cure for the evaluation of compositional distributional semantic models. In Proceedings of the Ninth International Confer- ence on Language Resources and Evaluation (LREC'14), pages 216-223, Reykjavik, Iceland, May 2014. European Language Resources Association (ELRA).","title":"A SICK cure for the evaluation of compositional distributional semantic models","authors":"Marco Marelli","year":"2014","venue":"Proceedings of the Ninth International Conference on Language Resources and Evaluation (LREC'14)","doi":null,"url":null,"internal_url":null},{"position":24,"raw":"[Muennighoff et al., 2022] Niklas Muennighoff, Nouamane Tazi, Loïc Magne, and Nils Reimers. Mteb: Massive text embedding benchmark. arXiv preprint arXiv:2210.07316, 2022.","title":"Mteb: Massive text embedding benchmark","authors":"Niklas Muennighoff, Nouamane Tazi, Loïc Magne, Nils Reimers","year":"2022","venue":null,"doi":null,"url":null,"internal_url":null},{"position":25,"raw":"[Resnik, 1995] Philip Resnik. Using information content to evaluate semantic similarity in a taxonomy. In Proceed- ings of the 14th International Joint Conference on Artifi- cial Intelligence -Volume 1, IJCAI'95, page 448-453, San Francisco, CA, USA, 1995.","title":"Using information content to evaluate semantic similarity in a taxonomy","authors":"Philip Resnik, Resnik","year":"1995","venue":"Proceedings of the 14th International Joint Conference on Artificial Intelligence","doi":null,"url":null,"internal_url":null},{"position":26,"raw":"Peter N. Robinson et al. In- terpretable clinical genomics with a likelihood ratio paradigm. The American Journal of Human Genetics, 107(3):403-417, 2020.","title":"Interpretable clinical genomics with a likelihood ratio paradigm","authors":"N Peter, Robinson","year":"2020","venue":"The American Journal of Human Genetics","doi":null,"url":null,"internal_url":null},{"position":27,"raw":"[Robinson et al., 2008] Peter N. Robinson, Sebastian Köhler, Sebastian Bauer, Dominik Seelow, Denise Horn, and Stefan Mundlos. The human phenotype ontology: A tool for annotating and analyzing human hereditary disease. The American Journal of Human Genetics, 83(5):610-615, 2008.","title":"The human phenotype ontology: A tool for annotating and analyzing human hereditary disease","authors":"Peter N Robinson, Sebastian Köhler, Sebastian Bauer, Dominik Seelow, Denise Horn, Stefan Mundlos","year":"2008","venue":"The American Journal of Human Genetics","doi":null,"url":null,"internal_url":null},{"position":28,"raw":"[Sogancıoglu et al., 2017] Gizem Sogancıoglu, Hakime Öztürk, and Arzucan Özgür. Biosses: a semantic sentence similarity estimation system for the biomedical domain. Bioinformatics, 33(14):i49-i58, 07 2017.","title":"Biosses: a semantic sentence similarity estimation system for the biomedical domain","authors":"Gizem Sogancıoglu, Hakime Öztürk, Arzucan Özgür","year":"2017","venue":"Bioinformatics","doi":null,"url":null,"internal_url":null},{"position":29,"raw":"[Wang et al., 2020] Wenhui Wang, Furu Wei, Li Dong, Hangbo Bao, Nan Yang, and Ming Zhou. Minilm: Deep self-attention distillation for task-agnostic compression of pre-trained transformers. In Advances in Neural Informa- tion Processing Systems, volume 33, pages 5776-5788, 2020.","title":"Minilm: Deep self-attention distillation for task-agnostic compression of pre-trained transformers","authors":"Wenhui Wang, Furu Wei, Li Dong, Hangbo Bao, Nan Yang, Ming Zhou","year":"2020","venue":"Advances in Neural Information Processing Systems","doi":null,"url":null,"internal_url":null},{"position":30,"raw":"[Wen et al., 2025] Baole Wen, Sheng Shi, Yi Long, Yanan Dang, and Weidong Tian. PhenoDP: leveraging deep learning for phenotype-based case reporting, disease rank- ing, and symptom recommendation. Genome Med., 17(1):67, June 2025.","title":"PhenoDP: leveraging deep learning for phenotype-based case reporting, disease ranking, and symptom recommendation","authors":"Baole Wen, Sheng Shi, Yi Long, Yanan Dang, Weidong Tian","year":"2025","venue":"Genome Med","doi":null,"url":null,"internal_url":null},{"position":31,"raw":"[Yuan et al., 2020] Zheng Yuan, Zhengyun Zhao, and Sheng Yu. CODER: knowledge infused cross-lingual med- ical term embedding for term normalization. CoRR, abs/2011.02947, 2020.","title":"CODER: knowledge infused cross-lingual medical term embedding for term normalization","authors":"Zheng Yuan, Zhengyun Zhao, Sheng Yu","year":"2020","venue":null,"doi":null,"url":null,"internal_url":null}],"license_note":"Copyright International Joint Conferences on Artificial Intelligence."}