Pith. sign in

REVIEW 2 major objections 6 minor 87 references

HypKG: Hypergraph-based Knowledge Graph Contextualization for Precision Healthcare

T0 review · 2 major / 6 minor · reviewed 2026-08-06 · deepseek-v4-flash

Challenge this review Re-run · record.json Download PDF Read on arXiv ↗

Pith's one-line read HypKG claims that patient-specific EHR context, added to a biomedical knowledge graph through a hypergraph, improves healthcare predictions by 12.15% on MIMIC-III and 9.66% on PROMOTE.

desk verdict Supervision confound undermines the headline gain, but the end-to-end hypergraph model is a real and fairly compared improvement over prior hypergraph baselines; worth refereeing with a request for a supervised KG baseline. read the letter →

arxiv 2507.19726 v2 pith:3MHYAM5N submitted 2025-07-26 cs.AI cs.LG

Yuzhang Xie , Xu Han , Ran Xu , Xiao Hu , Jiaying Lu , Carl Yang This is my paper · ORCID
classification cs.AIcs.LG
keywords contextualizedknowledgegraphhypergraphrepresentationprecisionhealthcareelectronichealthrecordsentitylinkingtransformerprediction
verification ladder T0 review T1 audit T2 compute T3 formal

The pith

A machine-rendered reading of the paper's core claim, the machinery that carries it, and where it could break.

The reading

General biomedical knowledge graphs store facts like "aspirin can prevent heart disease" but ignore the patient: the same fact can be wrong advice for someone with a bleeding disorder. HypKG claims that patient-specific context from electronic health records can be layered onto such a knowledge graph, and that doing so produces representations that predict health outcomes better than either source alone. The paper links each EHR medical attribute to a knowledge-graph entity, then builds one hypergraph (nodes are medical attributes, hyperedges are individual patient visits) and trains attention-based message passing on downstream prediction tasks. On MIMIC-III phenotyping and PROMOTE post-stroke cognitive impairment prediction, the authors report an average relative improvement of 12.15% and 9.66% over mean-pooled knowledge-graph embeddings across evaluation metrics. If this holds, precision healthcare can reuse large medical knowledge graphs already available, but with knowledge conditioned on each patient's actual record.

What carries the argument

The load-bearing object is the patient-context hypergraph: vertices are medical attributes (each linked to a KG entity) and hyperedges are patient visits, so a hyperedge can connect arbitrarily many attributes at once. On this hypergraph, HypKG runs a stack of Set Transformer-style message-passing layers with two alternating aggregators, $f_{V\to E}$ and $f_{E\to V}$: nodes within a hyperedge are pooled into a hyperedge embedding, then each node pools the hyperedges it belongs to, using multi-head attention (Eqs. 2--4). Initial node embeddings are 128-dimensional KG representations obtained with ComplEx on a subsampled iBKH graph, combined with graph-walk and language-model features and PCA-reduced, and the entire stack is supervised by downstream classification via binary cross-entropy.

What would settle it

Take the KG Embedding and Binary Embedding baselines, fine-tune them end-to-end with the same binary cross-entropy objective on the same training split (keeping the same hypergraph or classifier capacity), and compare test AUROC and AUCPR on MIMIC-III and PROMOTE; if the 12.15% and 9.66% gaps largely disappear, the reported gain comes from supervised fine-tuning rather than from hypergraph contextualization.

Watch

Extended reading notes

Core claim

HypKG's central claim is that contextualization should happen inside the knowledge representation itself: instead of embedding a knowledge graph once and pooling those vectors for a patient, or building a separate small graph per patient, the paper represents KG entities and patient visits in a single hypergraph and learns jointly. Nodes are EHR medical attributes linked to KG entities; each patient visit is a hyperedge connecting the attributes of that visit; and multi-head attention message passing repeatedly aggregates node information into hyperedges and hyperedge information back into nodes. Initial node features come from pretrained KG embeddings, and the whole hypergraph transformer is trained end-to-end with binary cross-entropy on the downstream healthcare prediction labels. The authors report that this consistently beats one-hot binary patient vectors and mean-pooled KG embeddings across accuracy, AUROC, AUCPR, and Macro-F1 on two real EHR datasets, and that entity pairs sharing patient contexts move closer in embedding space even when the KG has no direct relation between them.

Load-bearing premise

The comparison assumes the baselines and HypKG are equally informed by the downstream labels, but HypKG's hypergraph transformer is trained end-to-end with binary cross-entropy on the exact same prediction labels used for evaluation, while the BinaryEmbedding and KG Embedding baselines are not fine-tuned on those labels.

Editorial extensions

If this is right

  • If HypKG's gain is real, the same 128-dimensional contextualized patient representation can feed simple classifiers (LR, SVM, RF, XGB) and beat the much higher-dimensional one-hot encoding, which on MIMIC-III with SVM took over 10 hours—more than 20 times the time HypKG needed.
  • Contextualization improves the KG itself, not just the patient vectors: entity pairs that co-occur in patient contexts get larger similarity increases than pairs that never co-occur, so KG entity and relation embeddings are adjusted by patient data.
  • The framework tolerates imperfect entity linking: replacing PromptLink with BM25, BioBERT, or shuffled embeddings costs only 1–3% on most metrics, so deployment is possible in settings without a strong linker.
  • Joint modeling of EHR and KG is necessary for the reported result: ablations that remove either the KG initialization (EHR Only) or the real hyperedge structure (KG Only) fall well behind HypKG on both datasets.

Reading between the lines

Editorial extensions of the paper, not claims the author makes directly.

  • A consequence the paper leaves implicit is that the similarity-increase signal could be turned into a candidate generator: ranking entity pairs by how much their representations move together under patient context offers testable hypotheses for undiscovered drug-disease or comorbidity links.
  • The framework is indifferent to the source of the context, so a domain-specific entity linker plus a KG embedding initializer should transfer HypKG to e-commerce or user modeling; the authors flag this direction but do not test it.
  • The trained hypergraph itself is a reusable artifact: new patients can be slotted in as hyperedges over existing nodes and scored without re-embedding the whole KG, which would be a natural deployment-oriented extension.
Share X Bluesky LinkedIn Reddit HN

Editorial analysis

A structured set of objections, weighed in public.

Desk editor's note, referee report, and a circularity audit.

Referee Report

2 major / 6 minor

Summary. HypKG contextualizes a general biomedical knowledge graph with patient-specific EHR data by linking EHR attributes to KG entities via PromptLink, constructing a hypergraph whose nodes are medical attributes and whose hyperedges are patient visits, and training hypergraph transformers with a downstream classification loss. The paper evaluates the resulting embeddings and the end-to-end model on MIMIC-III phenotyping and PROMOTE post-stroke cognitive impairment, reporting average relative gains of 12.15% and 9.66% over a mean-pooled KG embedding baseline, together with ablations on entity linking, KG embedding initialization, joint modeling, hypergraph design, and hyperparameters.

Significance. The framework is well-motivated and the engineering contribution is real: it combines an LLM-based entity-linking pipeline with hypergraph transformers in a unified architecture, and it ships code and uses public data (MIMIC-III), which supports reproducibility. If the contextualization effect were cleanly isolated, the results would be of interest to the semantic-web and healthcare ML communities. However, the headline numerical claim is currently not supported as stated because the main comparison is confounded by unequal access to label supervision, so the expected significance is conditional on additional controlled experiments.

major comments (2)
  1. [Section 3.5 and Section 4.1, Table 2] The headline claim that HypKG achieves average relative gains of 12.15% on MIMIC-III and 9.66% on PROMOTE over the KG Embedding baseline is confounded by unequal supervision. The HypKG Embedding is produced by the model described in Section 3.5, which is trained with binary cross-entropy on the same downstream labels used for evaluation, whereas the KG Embedding baseline (Section 4.1) is a static mean-pooling of pretrained ComplEx embeddings with no label-based training, and BinaryEmbedding is a one-hot encoding. The comparison therefore conflates hypergraph contextualization with supervised fine-tuning on the evaluation labels. To support the causal claim, the authors should add a control condition in which the static KG embeddings are fine-tuned with the same BCE loss and optimizer (e.g., a fine-tuned MLP or direct embedding fine-tuning) and compare against that control in Table 2.
  2. [Supplemental Section G] The joint-modeling ablation does not remove the confound because all three arms, KG Only, EHR Only, and HypKG, train a hypergraph transformer with the same downstream BCE loss. The missing arm is a model initialized with the same KG node embeddings and trained with the same supervised loss but without hypergraph message passing/contextualization (e.g., a supervised MLP or linear classifier on the KG embeddings, or fine-tuned KG embeddings without the hypergraph structure). Without this arm, the performance gap attributed to contextualization could be explained entirely by supervised feature learning on the labels.
minor comments (6)
  1. [Section 4.2] The computation of the 'average relative improvement' is not defined; specify how the average is taken across metrics and datasets (e.g., mean of relative gains per metric), because the headline numbers are central.
  2. [Section 4.1 and Table 2] For the HypKG Embedding rows, clarify whether the extracted 128-D embeddings come from the epoch selected on validation, and whether the external ML models are trained on the same training split; this affects the interpretation of those rows.
  3. [Supplemental Table 4] The description of Z_lm,Xk says 'EHR medical attribute' but it should refer to a KG entity; correct the typo.
  4. [Table 3] Several 'Increase' values exceed 1 (e.g., 1.08, 1.21), which is only possible if the initial cosine similarities are negative; report the before/after similarities or state this explicitly, as the current text claims the increases 'approach 1'.
  5. [Section 3.3] Provide details of the GPT-4 prompt, candidate list construction, and any manual validation of linking accuracy; entity linking quality is a load-bearing component and the current description is high-level.
  6. [Supplemental Table 10] The omission of CompGCN + Large is attributed to runtime; report the hardware and time budget so readers can judge the completeness of the comparison.

Circularity Check

1 steps flagged · score 4.0 of 10

Main comparison is confounded: HypKG embeddings are trained with BCE on the same downstream labels used for evaluation, while the KG Embedding baseline is unsupervised mean-pooling; the reported 12.15%/9.66% gains are partly forced by this asymmetry, not by hypergraph contextualization alone.

  1. fitted input called prediction [Section 3.5 (Learning to Integrate); Section 4.1 (Baseline Methods); Section 4.2 (Table 2)]
    "Since we leverage supervised classification tasks for guidance, the final representation Zf inal is passed to a two-layer fully-connected layer (with size df inal=48) and the loss function is the binary cross-entropy. ... The resulting final representation, Zf inal, incorporates both patient-specific context and KG-derived knowledge ... KG Embedding: This method leverages KG information by mean-pooling KG representations based on the linked patient's medical attributes during a visit."

    HypKG's Table-2 embedding is Z_final from the hypergraph transformer trained with BCE on the same downstream task labels, while the KG Embedding baseline is mean-pooling of pretrained ComplEx embeddings with no label supervision and BinaryEmbedding is static one-hot. The reported average relative gains (12.15% MIMIC-III, 9.66% PROMOTE) therefore include the effect of supervised fine-tuning against the evaluation labels, a component absent from the baselines. Section G's KG Only and EHR Only ablations also train with the same BCE loss, so all arms share label supervision; no arm isolates hypergraph structure without label-based fitting. The headline comparison partly reduces, by the paper's own setup, to the asymmetry that only HypKG is optimized on downstream labels.

full rationale

The paper contains no first-principles derivation chain whose outputs are definitionally equivalent to its inputs; the method is an empirical hypergraph architecture initialized with KG embeddings and trained on downstream labels. The only substantial circularity is comparative: the central claim that contextualization improves KG representations is supported by comparing HypKG's label-trained Z_final against static, unsupervised KG Embedding and BinaryEmbedding baselines. Because the baselines never receive the downstream-task supervision that HypKG receives, the reported gains are not cleanly attributable to the hypergraph contextualization mechanism. Self-citations such as PromptLink and HypEHR are used as tools or baselines and are robustness-tested, so they are not load-bearing circularity by themselves. Overall, the design is partially circular in its attribution of the headline improvement, but the architecture itself has independent empirical content, giving a score of 4.

Assumptions & free parameters 6 free parameters · 3 assumptions · 0 invented entities

The central claim rests on several tuned hyperparameters and domain-specific modeling assumptions about entity linking and hypergraph construction, but it introduces no new physical or conceptual entities. The most consequential assumption is that supervision on downstream labels does not leak into the evaluation of the learned embeddings.

free parameters (6)
  • Learning rate = 1e-3 (MIMIC-III), 1e-4 (PROMOTE)
    Tuned per dataset; strongly affects optimization and final performance (Section D.2, Appendix I).
  • Hypergraph transformer layers L = 3
    Chosen by hyperparameter search; depth affects message passing (Section D.2).
  • Hidden state size d_hi = 48
    Tuned; controls model capacity (Section D.2).
  • Attention heads H = 4
    Tuned; affects attention granularity (Section D.2).
  • Relation filter threshold K = 800
    Subsampling threshold for KG embedding generation; chosen to balance scale and information (Section D.1).
  • Entity linking candidate count LC = 10
    Number of top candidates retrieved before LLM reranking (Section 3.3).
assumptions (3)
  • domain assumption EHR medical attributes can be accurately linked to KG entities via PromptLink (SAPBERT embeddings plus GPT-4 reasoning).
    The entire hypergraph construction depends on this linking step (Section 3.3). Ablations show robustness to linking errors, but accurate linking is still assumed for the main results.
  • domain assumption Patient visits as hyperedges and medical attributes as nodes capture clinically meaningful co-occurrence structure.
    This modeling choice determines what the hypergraph can represent (Section 3.4). If visit-level co-occurrence is not the right context, the method loses its justification.
  • domain assumption Pretrained KG embeddings (ComplEx, DeepWalk, SAPBERT) provide a useful initialization for node features.
    The method relies on these embeddings to inject KG knowledge; the joint-modeling ablation supports this, but it remains a modeling assumption (Sections 3.5, D.1, J).

how reviews work

0 comments
Cite this review

Pith. "Pith review of HypKG: Hypergraph-based Knowledge Graph Contextualization for Precision Healthcare." pith.science (2026). https://pith.science/paper/3MHYAM5N

@misc{pith2026250719726,
  author       = {Pith},
  title        = {Pith review of: HypKG: Hypergraph-based Knowledge Graph Contextualization for Precision Healthcare},
  year         = {2026},
  howpublished = {\url{https://pith.science/paper/3MHYAM5N}},
  note         = {Machine review of arXiv:2507.19726}
}
read the original abstract

Knowledge graphs (KGs) are important products of the semantic web, which are widely used in various application domains. Healthcare is one of such domains where KGs are intensively used, due to the high requirement for knowledge accuracy and interconnected nature of healthcare data. However, KGs storing general factual information often lack the ability to account for important contexts of the knowledge such as the status of specific patients, which are crucial in precision healthcare. Meanwhile, electronic health records (EHRs) provide rich personal data, including various diagnoses and medications, which provide natural contexts for general KGs. In this paper, we propose HypKG, a framework that integrates patient information from EHRs into KGs to generate contextualized knowledge representations for accurate healthcare predictions. Using advanced entity-linking techniques, we connect relevant knowledge from general KGs with patient information from EHRs, and then utilize a hypergraph model to "contextualize" the knowledge with the patient information. Finally, we employ hypergraph transformers guided by downstream prediction tasks to jointly learn proper contextualized representations for both KGs and patients, fully leveraging existing knowledge in KGs and patient contexts in EHRs. In experiments using a large biomedical KG and two real-world EHR datasets, HypKG demonstrates significant improvements in healthcare prediction tasks across multiple evaluation metrics. Additionally, by integrating external contexts, HypKG can learn to adjust the representations of entities and relations in KG, potentially improving the quality and real-world utility of knowledge.

Figures

Figures reproduced from arXiv: 2507.19726 by the authors.

Figure 1
Figure 1. A toy example of KG contextualization. Left: traditional KG. Right: our pro￾posed contextualized KG. the most relevant knowledge from large-scale KGs, tailoring it to the individual patient’s specific needs. This integration of EHR and KG enriches the existing knowledge with patient-specific contexts, which could serve as a foundation for a wide range of downstream healthcare applications. To combine KG knowledge wi… view at source ↗
Figure 2
Figure 2. Overview of HypKG framework. Left: Linking knowledge and contexts. Middle: Jointly modeling knowledge (e.g., node V1) and contexts (e.g., hyperedge E1, which includes nodes V1 and V2; hyperedge E2, which includes nodes V2, V3, and V4). Right: Learning to integrate knowledge and contexts [PITH_FULL_IMAGE:figures/full_fig_p007_2.png] view at source ↗
Figure 3
Figure 3. Histogram showing the similarity increase for two different classes after HypKG’s learning process in the MIMIC-III dataset [PITH_FULL_IMAGE:figures/full_fig_p033_3.png] view at source ↗
Figures from the paper (1 more)
Figure 4
Figure 4. Figure 4: Histogram showing the similarity increase for two different classes after HypKG’s learning process in the PROMOTE dataset [PITH_FULL_IMAGE:figures/full_fig_p033_4.png]

Discussion (0). Continue with ORCID to comment.

Reference graph

Works this paper leans on

87 extracted references · 70 canonical work pages

  1. [1]

    Journal of Big Data10(1), 81 (2023)

    Abu-Salih, B., Al-Qurishi, M., Alweshah, M., Al-Smadi, M., Alfayez, R., Saadeh, H.: Healthcare knowledge graph construction: A systematic review of the state-of- the-art, open issues, and opportunities. Journal of Big Data10(1), 81 (2023)

    2023

  2. [2]

    Cell177(1), 58–69 (2019)

    Abul-Husn, N.S., Kenny, E.E.: Personalized medicine and the power of electronic health records. Cell177(1), 58–69 (2019)

    2019

  3. [3]

    Snomed2Vec: Random Walk and Poincar\'e Embeddings of a Clinical Knowledge Base for Healthcare Analytics

    Agarwal, K., Eftimov, T., Addanki, R., Choudhury, S., Tamang, S., Rallo, R.: Snomed2vec: Random walk and poincare embeddings of a clinical knowledge base for healthcare analytics. arXiv preprint arXiv:1907.08650 (2019)

    work page Pith review arXiv 2019

  4. [4]

    Journal of Machine Learning Research22(82), 1–6 (2021)

    Ali, M., Berrendorf, M., Hoyt, C.T., Vermue, L., Sharifzadeh, S., Tresp, V., Lehmann, J.: Pykeen 1.0: a python library for training and evaluating knowledge graph embeddings. Journal of Machine Learning Research22(82), 1–6 (2021)

    2021

  5. [5]

    Journal of clinical orthopaedics and trauma 13, 53–56 (2021)

    Amer, K.M., Congiusta, D.V., Suri, P., Choudhry, A., Otero, K., Adams, M.: Clav- icle fractures: Associated trauma and morbidity. Journal of clinical orthopaedics and trauma 13, 53–56 (2021)

    2021

  6. [6]

    Journal of the American Medical Informatics Association17(3), 229–236 (2010)

    Aronson, A.R., Lang, F.M.: An overview of metamap: historical perspective and recent advances. Journal of the American Medical Informatics Association17(3), 229–236 (2010)

    2010

  7. [7]

    Pattern Recognition 110, 107637 (2021)

    Bai, S., Zhang, F., Torr, P.H.: Hypergraph convolution and hypergraph attention. Pattern Recognition 110, 107637 (2021)

    2021

  8. [8]

    npj Digital Medicine8(1), 166 (2025)

    Bhasuran, B., Jin, Q., Xie, Y., Yang, C., Hanna, K., Costa, J., Shavor, C., Han, W., Lu, Z., He, Z.: Preliminary analysis of the impact of lab results on large language model generated differential diagnoses. npj Digital Medicine8(1), 166 (2025)

    2025

Show all 87 references
  1. [9]

    Nucleic acids research32(suppl_1), D267–D270 (2004)

    Bodenreider, O.: The unified medical language system (umls): integrating biomed- ical terminology. Nucleic acids research32(suppl_1), D267–D270 (2004)

    2004

  2. [10]

    In: Proceedings of the AAAI conference on artificial intelli- gence

    Bordes, A., Weston, J., Collobert, R., Bengio, Y.: Learning structured embeddings of knowledge bases. In: Proceedings of the AAAI conference on artificial intelli- gence. vol. 25, pp. 301–306 (2011)

    2011

  3. [11]

    ACM Computing Surveys 56(6), 1–42 (2024)

    Cao, J., Fang, J., Meng, Z., Liang, S.: Knowledge graph embedding: A survey from the perspective of representation spaces. ACM Computing Surveys 56(6), 1–42 (2024)

    2024

  4. [12]

    BMC Medical Informatics and Decision Making23(1), 12 (2023)

    Carvalho, R.M., Oliveira, D., Pesquita, C.: Knowledge graph embeddings for icu readmission prediction. BMC Medical Informatics and Decision Making23(1), 12 (2023)

    2023

  5. [13]

    R version90(1- 66), 40 (2019)

    Chen, T., He, T., Benesty, M., Khotilovich, V.: Package ‘xgboost’. R version90(1- 66), 40 (2019)

    2019

  6. [14]

    Chien, E., Pan, C., Peng, J., Milenkovic, O.: You are allset: A multiset function frameworkforhypergraphneuralnetworks.arXivpreprintarXiv:2106.13264(2021)

    2021 arXiv

  7. [15]

    AMIA Summits on Translational Science Proceedings2016, 41 (2016)

    Choi, Y., Chiu, C.Y.I., Sontag, D.: Learning low-dimensional representations of medical concepts. AMIA Summits on Translational Science Proceedings2016, 41 (2016)

    2016

  8. [16]

    Best Practice & Research Clinical Rheumatology29(1), 6–19 (2015)

    Clauw, D.J.: Diagnosing and treating chronic musculoskeletal pain based on the underlying mechanism (s). Best Practice & Research Clinical Rheumatology29(1), 6–19 (2015)

    2015

  9. [17]

    Journal of Biomedical Informatics p

    Cui, H., Lu, J., Xu, R., Wang, S., Ma, W., Yu, Y., Yu, S., Kan, X., Ling, C., Zhao, L., Qin, Z.S., Ho, J.C., Fu, T., Ma, J., Huai, M., Wang, F., Yang, C.: A review on knowledge graphs for healthcare: Resources, applications, and promises. Journal of Biomedical Informatics p. 1...

    2025

  10. [18]

    International journal of nephrology and renovascular disease pp

    Dalal, P., Shah, G., Chhabra, D., Gallon, L.: Role of tacrolimus combination ther- apy with mycophenolate mofetil in the prevention of organ rejection in kidney transplant patients. International journal of nephrology and renovascular disease pp. 107–115 (2010)

    2010

  11. [19]

    Jama307(21), 2286–2294 (2012)

    De Berardis, G., Lucisano, G., D’Ettorre, A., Pellegrini, F., Lepore, V., Tognoni, G., Nicolucci, A.: Association of aspirin use with major bleeding in patients with and without diabetes. Jama307(21), 2286–2294 (2012)

    2012

  12. [20]

    D’Souza, J., Ng, V.: Sieve-based entity linking for the biomedical domain. In: Pro- ceedings of the 53rd Annual Meeting of the Association for Computational Linguis- tics and the 7th International Joint Conference on Natural Language Processing (Volume 2: Short Papers). pp. 29...

    2015

  13. [21]

    Blood, The Journal of the Amer- ican Society of Hematology140(6), 571–593 (2022)

    Gallagher, P.G.: Anemia in the pediatric patient. Blood, The Journal of the Amer- ican Society of Hematology140(6), 571–593 (2022)

    2022

  14. [22]

    arXiv preprint arXiv:2308.14321 (2023)

    Gao, Y., Li, R., Caskey, J., Dligach, D., Miller, T., Churpek, M.M., Afshar, M.: Leveraging a medical knowledge graph into large language models for diagnosis prediction. arXiv preprint arXiv:2308.14321 (2023)

    2023 arXiv

  15. [23]

    Journal of the American College of Cardiology 51(19), 1829–1843 (2008)

    Gasparyan, A.Y., Watson, T., Lip, G.Y.: The role of aspirin in cardiovascular prevention: implications of aspirin resistance. Journal of the American College of Cardiology 51(19), 1829–1843 (2008)

    2008

  16. [24]

    Transfusion medicine and hemotherapy43(3), 203– 222 (2016)

    German Advisory Committee Blood (Arbeitskreis Blut), S.A.o.P.T.b.B.: Human immunodeficiency virus (hiv). Transfusion medicine and hemotherapy43(3), 203– 222 (2016)

    2016

  17. [25]

    Journal of pain research pp

    Hulens, M., Rasschaert, R., Vansant, G., Stalmans, I., Bruyninckx, F., Dankaerts, W.: The link between idiopathic intracranial hypertension, fibromyalgia, and chronic fatigue syndrome: exploration of a shared pathophysiology. Journal of pain research pp. 3129–3140 (2018)

    2018

  18. [26]

    87–104 (2021)

    Imambi,S.,Prakash,K.B.,Kanagachidambaresan,G.:Pytorch.Programmingwith TensorFlow: solution for edge computing applications pp. 87–104 (2021)

    2021

  19. [27]

    AMIA Summits on Translational Science Proceedings2020, 269 (2020)

    Ji, Z., Wei, Q., Xu, H.: Bert-based ranking for biomedical entity normalization. AMIA Summits on Translational Science Proceedings2020, 269 (2020)

    2020

  20. [28]

    In: The Twelfth International Confer- ence on Learning Representations (2024)

    Jiang, P., Xiao, C., Cross, A.R., Sun, J.: Graphcare: Enhancing healthcare predic- tions with personalized knowledge graphs. In: The Twelfth International Confer- ence on Learning Representations (2024)

    2024

  21. [29]

    Scientific data3(1), 1–9 (2016)

    Johnson, A.E., Pollard, T.J., Shen, L., Lehman, L.w.H., Feng, M., Ghassemi, M., Moody,B.,Szolovits,P.,AnthonyCeli,L.,Mark,R.G.:Mimic-iii,afreelyaccessible critical care database. Scientific data3(1), 1–9 (2016)

    2016

  22. [30]

    Annual Review of Biomedical Data Science7 (2024)

    Johnson, R., Li, M.M., Noori, A., Queen, O., Zitnik, M.: Graph artificial intelli- gence in medicine. Annual Review of Biomedical Data Science7 (2024)

    2024

  23. [31]

    Primary Care Diabetes15(5), 848–852 (2021)

    Jońca, M., Krótki, F., Tomasik, P.: The effect of disinfecting procedure on the glu- cose concentration using a personal glucose meter. Primary Care Diabetes15(5), 848–852 (2021)

    2021

  24. [32]

    Information 13(4), 161 (2022)

    Kejriwal, M.: Knowledge graphs: A practical review of the research landscape. Information 13(4), 161 (2022)

    2022

  25. [33]

    In: 3rd Inter- national Conference on Learning Representations (2015)

    Kingma, D.P., Ba, J.: Adam: A method for stochastic optimization. In: 3rd Inter- national Conference on Learning Representations (2015)

    2015

  26. [34]

    Journal of medical Internet research23(3), e22219 (2021) HypKG: Hypergraph KG Contextualization for Healthcare 19

    Kohane, I.S., Aronow, B.J., Avillach, P., Beaulieu-Jones, B.K., Bellazzi, R., Brad- ford, R.L., Brat, G.A., Cannataro, M., Cimino, J.J., García-Barrio, N., et al.: What every reader shouldknowabout studiesusing electronic healthrecord databut may be afraid to ask. Journal of m...

    2021

  27. [35]

    Case reports in medicine 2012(1), 649850 (2012)

    Kurugol, Z., Onen, S.S., Koturoglu, G.: Severe hemolytic anemia associated with mild pneumonia caused by mycoplasma pneumonia. Case reports in medicine 2012(1), 649850 (2012)

    2012

  28. [36]

    arXiv preprint arXiv:2401.08878 (2024)

    Lee,G.,Bu,F.,Eliassi-Rad,T.,Shin,K.:Asurveyonhypergraphmining:Patterns, tools, and generators. arXiv preprint arXiv:2401.08878 (2024)

    2024 arXiv

  29. [37]

    Bioinformatics 36(4), 1234–1240 (2020)

    Lee, J., Yoon, W., Kim, S., Kim, D., Kim, S., So, C.H., Kang, J.: Biobert: a pre-trained biomedical language representation model for biomedical text mining. Bioinformatics 36(4), 1234–1240 (2020)

    2020

  30. [38]

    In: Interna- tional conference on machine learning

    Lee, J., Lee, Y., Kim, J., Kosiorek, A., Choi, S., Teh, Y.W.: Set transformer: A framework for attention-based permutation-invariant neural networks. In: Interna- tional conference on machine learning. pp. 3744–3753. PMLR (2019)

    2019

  31. [39]

    ACM Transactions on Knowledge Discovery from Data17(5), 1–22 (2023)

    Li, M., Zhang, Y., Li, X., Zhang, Y., Yin, B.: Hypergraph transformer neural net- works. ACM Transactions on Knowledge Discovery from Data17(5), 1–22 (2023)

    2023

  32. [40]

    arXiv preprint arXiv:2010.11784 (2020)

    Liu, F., Shareghi, E., Meng, Z., Basaldella, M., Collier, N.: Self-alignment pretrain- ing for biomedical entity representations. arXiv preprint arXiv:2010.11784 (2020)

    2020 arXiv

  33. [41]

    arXiv preprint arXiv:2503.23167 (2025)

    Liu, Z., Wang, X., Wang, B., Huang, Z., Yang, C., Jin, W.: Graph odes and beyond: A comprehensive survey on integrating differential equations with graph neural networks. arXiv preprint arXiv:2503.23167 (2025)

    2025 arXiv

  34. [42]

    In: 46th Interna- tional ACM SIGIR Conference on Research and Development in Information Re- trieval - Short Paper (2023)

    Lu, J., Shen, J., Xiong, B., Ma, W., Staab, S., Yang, C.: Hiprompt: Few-shot biomedical knowledge fusion via hierarchy-oriented prompting. In: 46th Interna- tional ACM SIGIR Conference on Research and Development in Information Re- trieval - Short Paper (2023)

    2023

  35. [43]

    Headache: The Journal of Head and Face Pain56(2), 436–446 (2016)

    Lucas, S.: The pharmacology of indomethacin. Headache: The Journal of Head and Face Pain56(2), 436–446 (2016)

    2016

  36. [44]

    Diabetes Care38(10), e158–e159 (2015)

    Maahs, D.M., DeSalvo, D., Pyle, L., Ly, T., Messer, L., Clinton, P., Westfall, E., Wadwa, R.P., Buckingham, B.: Effect of acetaminophen on cgm glucose in an outpatient setting. Diabetes Care38(10), e158–e159 (2015)

    2015

  37. [45]

    Measurement: Sensors 25, 100577 (2023)

    Manikandan, R., Kuwelkar, S., Sivakumar, R.: An hybrid technique for optimized clustering of ehr using binary particle swarm and constrained optimization for better performance in prediction of cardiovascular diseases. Measurement: Sensors 25, 100577 (2023)

    2023

  38. [46]

    Journal of biomedical informatics143, 104403 (2023)

    Murali, L., Gopakumar, G., Viswanathan, D.M., Nedungadi, P.: Towards electronic health record-based medical knowledge graph construction, completion, and appli- cations: A literature study. Journal of biomedical informatics143, 104403 (2023)

    2023

  39. [47]

    Radio- graphics 41(4), 1123–1143 (2021)

    Naeem, M., Zulfiqar, M., Siddiqui, M.A., Shetty, A.S., Haq, A., Varela, C., Siegel, C., Menias, C.O.: Imaging manifestations of genitourinary tuberculosis. Radio- graphics 41(4), 1123–1143 (2021)

    2021

  40. [48]

    The Journal of Pediatric Pharmacology and Therapeutics 27(6), 524–528 (2022)

    Novick, K., Cober, M.P.: Evaluation of inpatient starter parenteral nutrition use in the neonatal intensive care unit. The Journal of Pediatric Pharmacology and Therapeutics 27(6), 524–528 (2022)

    2022

  41. [49]

    Artificial Intelligence in Medicine131, 102359 (2022)

    Ochoa, J.G.D., Mustafa, F.E.: Graph neural network modelling as a potentially effective method for predicting and analyzing procedures based on patients’ diag- noses. Artificial Intelligence in Medicine131, 102359 (2022)

    2022

  42. [50]

    the Journal of machine Learning research12, 2825–2830 (2011)

    Pedregosa, F., Varoquaux, G., Gramfort, A., Michel, V., Thirion, B., Grisel, O., Blondel, M., Prettenhofer, P., Weiss, R., Dubourg, V., et al.: Scikit-learn: Machine learning in python. the Journal of machine Learning research12, 2825–2830 (2011)

    2011

  43. [51]

    AIDS research and human retroviruses29(2), 256–265 (2013) 20 Y

    Reynes, J., Trinh, R., Pulido, F., Soto-Malave, R., Gathe, J., Qaqish, R., Tian, M., Fredrick, L., Podsadecki, T., Norton, M., et al.: Lopinavir/ritonavir combined with raltegravir or tenofovir/emtricitabine in antiretroviral-naive subjects: 96-week results of the progress stu...

    2013

  44. [52]

    Foundations and Trends® in Information Retrieval 3(4), 333–389 (2009)

    Robertson, S., Zaragoza, H., et al.: The probabilistic relevance framework: Bm25 and beyond. Foundations and Trends® in Information Retrieval 3(4), 333–389 (2009)

    2009

  45. [53]

    Jama291(3), 325–334 (2004)

    Samore, M.H., Evans, R.S., Lassen, A., Gould, P., Lloyd, J., Gardner, R.M., Abouzelof,R.,Taylor,C.,Woodbury,D.A.,Willy,M.,etal.:Surveillanceofmedical device–related hazards and adverse events in hospitalized patients. Jama291(3), 325–334 (2004)

    2004

  46. [54]

    European Journal of Paediatric Neurology (2025)

    Schüssler,S.C.,Paul,A.,Niederreiter,U.,Deiters,L.,Fahlbusch,F.B.,Morhart,P., Trollmann, R.: Seizures in preterm infants with germinal-matrix-intraventricular hemorrhage (gm-ivh): a retrospective monocentric study on predictors and neu- rodevelopmental outcome. European Journal...

    2025

  47. [55]

    World Journal of Critical Care Medicine12(3), 130 (2023)

    Singh, O., Juneja, D.: Upper extremity deep vein thrombosis: an intensivist’s per- spective. World Journal of Critical Care Medicine12(3), 130 (2023)

    2023

  48. [56]

    Nature 620(7972), 172–180 (2023)

    Singhal, K., Azizi, S., Tu, T., Mahdavi, S.S., Wei, J., Chung, H.W., Scales, N., Tanwani, A., Cole-Lewis, H., Pfohl, S., et al.: Large language models encode clinical knowledge. Nature 620(7972), 172–180 (2023)

    2023

  49. [57]

    Speer, R., Chin, J., Havasi, C.: Conceptnet 5.5: An open multilingual graph of generalknowledge.In:ProceedingsoftheAAAIconferenceonartificialintelligence. vol. 31 (2017)

    2017

  50. [58]

    Iscience26(4) (2023)

    Su, C., Hou, Y., Zhou, M., Rajendran, S., Maasch, J.R., Abedi, Z., Zhang, H., Bai, Z., Cuturrufo, A., Guo, W., et al.: Biomedical discovery through the integrative biomedical knowledge hub (ibkh). Iscience26(4) (2023)

    2023

  51. [59]

    Journal of healthcare engineering 2018 (2018)

    Sun, W., Cai, Z., Li, Y., Liu, F., Fang, S., Wang, G.: Data processing and text mining technologies on electronic medical records: a review. Journal of healthcare engineering 2018 (2018)

    2018

  52. [60]

    In: Proceedings of the 12th Knowledge Capture Conference 2023

    Theodoropoulos, C., Mulligan, N., Stappenbeck, T., Bettencourt-Silva, J.: Repre- sentation learning for person or entity-centric knowledge graphs: An application in healthcare. In: Proceedings of the 12th Knowledge Capture Conference 2023. pp. 225–233 (2023)

    2023

  53. [61]

    In: International conference on machine learning

    Trouillon, T., Welbl, J., Riedel, S., Gaussier, É., Bouchard, G.: Complex embed- dings for simple link prediction. In: International conference on machine learning. pp. 2071–2080. PMLR (2016)

    2016

  54. [62]

    In: International Conference on Learning Representations

    Vashishth, S., Sanyal, S., Nitin, V., Talukdar, P.: Composition-based multi- relational graph convolutional networks. In: International Conference on Learning Representations

  55. [63]

    Journal of Computational Biology 24(8), 767–786 (2017)

    Vasiljeva, I., Arandjelović, O.: Diagnosis prediction from electronic health records using the binary diagnosis history vector representation. Journal of Computational Biology 24(8), 767–786 (2017)

    2017

  56. [64]

    American journal of health- system pharmacy 55(2), 154–158 (1998)

    Walker, S.E., Gray, S., Schmidt, B.: Stability of reconstituted indomethacin sodium trihydrate in original vials and polypropylene syringes. American journal of health- system pharmacy 55(2), 154–158 (1998)

    1998

  57. [65]

    ACM Computing Sur- veys 56(3), 1–52 (2023)

    Wang, B., Xie, Q., Pei, J., Chen, Z., Tiwari, P., Li, Z., Fu, J.: Pre-trained lan- guage models in biomedical domain: A systematic survey. ACM Computing Sur- veys 56(3), 1–52 (2023)

    2023

  58. [66]

    Stroke 53(4), 1054–1064 (2022)

    Webb, A.J., Werring, D.J.: New insights into cerebrovascular pathophysiology and hypertension. Stroke 53(4), 1054–1064 (2022)

    2022

  59. [67]

    Wolff, J., Reißner, P., Hefner, G., Normann, C., Kaier, K., Binder, H., Hiemke, C., Toto, S., Domschke, K., Marschollek, M., et al.: Pharmacotherapy, drug-drug interactionsandpotentiallyinappropriatemedicationindepressivedisorders.PLoS One 16(7), e0255192 (2021) HypKG: Hypergr...

    2021

  60. [68]

    Medical care48(6), S106–S113 (2010)

    Wu, J., Roy, J., Stewart, W.F.: Prediction modeling using ehr data: challenges, strategies, and a comparison of machine learning approaches. Medical care48(6), S106–S113 (2010)

    2010

  61. [69]

    In: STROKE

    Wu, Y., Nahab, F., Ge, Y., Xie, Y., Aboul-Nour, H., Yang, C., Hu, X.: Prediction of post-stroke af in esus patients is enhanced by combining expert-derived predic- tors and embedding of full diagnostic codes using pre-trained hypergraph neural networks. In: STROKE. vol. 56. LI...

    2025

  62. [70]

    arXiv preprint arXiv:2507.02773 (2025)

    Xie, Y., Cui, H., Zhang, Z., Lu, J., Shu, K., Nahab, F., Hu, X., Yang, C.: Kerap: A knowledge-enhancedreasoningapproachforaccuratezero-shotdiagnosisprediction using multi-agent llms. arXiv preprint arXiv:2507.02773 (2025)

    2025 arXiv

  63. [71]

    In: Proceedings of the 47th International ACM SIGIR Conference on Research and Development in Information Retrieval

    Xie, Y., Lu, J., Ho, J., Nahab, F., Hu, X., Yang, C.: Promptlink: Leveraging large language models for cross-source biomedical concept linking. In: Proceedings of the 47th International ACM SIGIR Conference on Research and Development in Information Retrieval. pp. 2589–2593 (2024)

    2024

  64. [72]

    Stroke 56(Suppl_1), AWP175–AWP175 (2025)

    Xie, Y., Nahab, F., Ge, Y., Wu, Y., Saurman, J., Yang, C., Hu, X.: Abstract wp175: Predicting post-stroke cognitive impairment (psci) using multiple machine learning approaches. Stroke 56(Suppl_1), AWP175–AWP175 (2025)

    2025

  65. [73]

    In: 2022 IEEE international conference on bioinformatics and biomedicine (BIBM)

    Xie, Y., Niu, G., Da, Q., Dai, W., Yang, Y.: Survival prediction for gastric cancer via multimodal learning of whole slide images and gene expression. In: 2022 IEEE international conference on bioinformatics and biomedicine (BIBM). pp. 1311–

    2022

  66. [74]

    Artificial Intelligence in Medicine152, 102871 (2024)

    Xie, Y., Sang, Q., Da, Q., Niu, G., Deng, S., Feng, H., Chen, Y., Li, Y.Y., Liu, B., Yang, Y., et al.: Improving diagnosis and outcome prediction of gastric cancer via multimodal learning using whole slide pathological images and gene expression. Artificial Intelligence in Med...

    2024

  67. [75]

    In: Proceedings of the 58th Annual Meeting of the Association for Computational Linguistics

    Xu, D., Zhang, Z., Bethard, S.: A generate-and-rank framework with semantic type regularization for biomedical concept normalization. In: Proceedings of the 58th Annual Meeting of the Association for Computational Linguistics. pp. 8452–8464 (2020)

    2020

  68. [76]

    AMIA Summits on Translational Science Proceedings2023, 582 (2023)

    Xu, R., Ali, M.K., Ho, J.C., Yang, C.: Hypergraph transformers for ehr-based clinical predictions. AMIA Summits on Translational Science Proceedings2023, 582 (2023)

    2023

  69. [77]

    In: Machine Learning for Health

    Xu, R., Yu, Y., Zhang, C., Ali, M.K., Ho, J.C., Yang, C.: Counterfactual and factual reasoning over hypergraphs for interpretable clinical predictions on ehr. In: Machine Learning for Health. pp. 259–278. PMLR (2022)

    2022

  70. [78]

    In: Proceedings of the ACM Web Conference 2023

    Xu, Y., Chu, X., Yang, K., Wang, Z., Zou, P., Ding, H., Zhao, J., Wang, Y., Xie, B.: Seqcare: Sequential training with external medical knowledge graph for diagnosis prediction in healthcare data. In: Proceedings of the ACM Web Conference 2023. pp. 2819–2830 (2023)

    2023

  71. [79]

    Advances in neural information processing systems32 (2019)

    Yadati, N., Nimishakavi, M., Yadav, P., Nitin, V., Louis, A., Talukdar, P.: Hy- pergcn: A new method for training graph convolutional networks on hypergraphs. Advances in neural information processing systems32 (2019)

    2019

  72. [80]

    Advances in Neural Information Processing Systems35, 8374–8385 (2022)

    Yang, H., Lin, Z., Zhang, M.: Rethinking knowledge graph evaluation under the open-world assumption. Advances in Neural Information Processing Systems35, 8374–8385 (2022)

    2022

  73. [81]

    In: Proceedings of the 45th international ACM SIGIR conference on research and development in information retrieval

    Yang, Y., Huang, C., Xia, L., Li, C.: Knowledge graph contrastive learning for recommendation. In: Proceedings of the 45th international ACM SIGIR conference on research and development in information retrieval. pp. 1434–1443 (2022) 22 Y. Xie et al

    2022

  74. [82]

    Ye, M., Cui, S., Wang, Y., Luo, J., Xiao, C., Ma, F.: Medpath: Augmenting health riskpredictionviamedicalknowledgepaths.In:ProceedingsoftheWebConference

  75. [83]

    In: Proceedings of the 2021 Conference of the North American Chapter of the Association for Computational Linguistics: Human Language Technologies

    Yu, Y., Zuo, S., Jiang, H., Ren, W., Zhao, T., Zhang, C.: Fine-tuning pre-trained language model with weak supervision: A contrastive-regularized self-training ap- proach. In: Proceedings of the 2021 Conference of the North American Chapter of the Association for Computational...

    2021

  76. [84]

    In: Findings of the Association for Computational Linguistics: EMNLP 2022

    Zhang, S., Cheng, H., Vashishth, S., Wong, C., Xiao, J., Liu, X., Naumann, T., Gao, J., Poon, H.: Knowledge-rich self-supervision for biomedical entity linking. In: Findings of the Association for Computational Linguistics: EMNLP 2022. pp. 868–880 (2022)

    2022

  77. [85]

    Information Pro- cessing & Management57(6), 102324 (2020)

    Zhang, Y., Sheng, M., Zhou, R., Wang, Y., Han, G., Zhang, H., Xing, C., Dong, J.: Hkgb: an inclusive, extensible, intelligent, semi-auto-constructed knowledge graph framework for healthcare with clinicians’ expertise incorporated. Information Pro- cessing & Management57(6), 10...

    2020

  78. [86]

    Preva- lence

    Zhang, Z., Cui, H., Xu, R., Xie, Y., Ho, J.C., Yang, C.: Tacco: Task-guided co- clustering of clinical concepts and patient visits for disease subtyping based on ehr data. In: Proceedings of the 30th ACM SIGKDD Conference on Knowledge Discovery and Data Mining. pp. 6324–6334 (...

    2024

  79. [2021]

    1397–1409 (2021)

    pp. 1397–1409 (2021)

    2021

Pith tools

Reviewed August 6, 2026 · model on record in the stance chip above.