pith. sign in

arxiv: 2606.07677 · v1 · pith:EZCBF3A4new · submitted 2026-06-04 · 📊 stat.ML · cs.LG· stat.AP· stat.ME

Disentangling Latent Risk Pathways via Bayesian Hypergraph Inference

Shengxian Ding , Haonan Gao , Pangpang Liu , Xinyuan Tian , Yize Zhao This is my paper

Pith reviewed 2026-06-27 23:06 UTC · model grok-4.3

classification 📊 stat.ML cs.LGstat.APstat.ME
keywords bayesian hypergraphdisease pathwaysEHR modelingrisk factorslatent structurevariational inferencemulti-diseaseuncertainty quantification
0
0 comments X

The pith

Bayesian hypergraph inference models latent disease pathways modulated by risk factors using EHR data.

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

The paper presents a Bayesian hypergraph inference framework to model how risk factors influence groups of diseases in electronic health records. It treats hyperedges as latent subsets of diseases that share risk patterns, letting individual diseases belong to several such groups. This setup aims to reveal higher-order structures and offer uncertainty estimates, improving on independent or black-box approaches especially for rare diseases.

Core claim

The Bayesian hypergraph inference framework reframes multi-disease modeling around latent, risk-factor-modulated disease pathways. Risk factors act on hyperedges, which are latent disease subsets with shared risk patterns, allowing diseases to participate in multiple distinct pathways and enabling interpretable, higher-order structure beyond pairwise associations. A repulsion prior encourages parsimonious and identifiable structure, while posterior inference provides calibrated uncertainty over both disease groupings and risk-factor influence.

What carries the argument

Bayesian hypergraph inference framework with repulsion prior and structured variational inference, where hyperedges represent latent disease subsets modulated by risk factors.

If this is right

  • Diseases participate in multiple pathways, enabling higher-order structure beyond pairwise associations.
  • Improved estimation for rare diseases through shared pathway information.
  • Calibrated posterior uncertainty over disease groupings and risk-factor effects.
  • Scalable inference on large EHR datasets via structured variational methods.
  • Stable and interpretable pathway structure on both simulated and real data.

Where Pith is reading between the lines

These are editorial extensions of the paper, not claims the author makes directly.

  • The approach could generalize to other multi-outcome settings with shared latent factors, such as species co-occurrence in ecology.
  • One could test whether the inferred pathways recover known biological or clinical groupings when applied to independent cohorts.
  • The repulsion prior might offer advantages in identifiability for other Bayesian network models with overlapping groups.

Load-bearing premise

The repulsion prior combined with the structured variational inference will yield parsimonious, identifiable hyperedge structures on real EHR data without extensive post-hoc tuning.

What would settle it

If the model applied to UK Biobank data produces hyperedges whose risk factor associations do not match established medical knowledge or if uncertainty calibration fails in cross-validation for rare outcomes.

Figures

Figures reproduced from arXiv: 2606.07677 by Haonan Gao, Pangpang Liu, Shengxian Ding, Xinyuan Tian, Yize Zhao.

Figure 1
Figure 1. Figure 1: Bayesian Hypergraph Pathway Inference Workflow. (a) Input data consist of patient covariates and multiple disease outcomes. (b) A latent disease hypergraph models higher-order structure via hyperedges, allowing diseases to participate in multiple pathways with uncertain existence and membership. (c) Risk factors act on hyperedges to induce structured effects across diseases, yielding disentangled attributi… view at source ↗
Figure 2
Figure 2. Figure 2: Predictive stability via structured information sharing. Distribution of per-disease ∆AUC scores (relative to tuned Logis￾tic Regression) stratified by disease prevalence. While baselines like LGBM and Classifier Chains exhibit high variance and insta￾bility for rare diseases, BHPI maintains robust performance by borrowing strength across shared latent pathways. sion probabilities (PIPs) and reduces redund… view at source ↗
Figure 3
Figure 3. Figure 3: Disentanglement and parsimony of risk-modulated pathways. (Left) PIPs for disease–pathway membership and risk-factor activation, revealing sparse, overlapping disease organization and pathway-specific modulation with quantified uncertainty. (Right) Summary of the thresholded structure (PIP(ze = 1) > 0.5), showing compact pathway sizes, limited pathway participation per disease and risk factor, and overall … view at source ↗
Figure 4
Figure 4. Figure 4: Global visualization of the learned risk-factor–to-disease pathway structure (hyperedges with PIP(ze = 1) > 0.5) (Left). Risk factors (top) connect to latent pathways (middle), which in turn connect to diseases (bottom). Edge widths correspond to PIPs, illustrating uncertainty, overlap, and modular risk propagation. Node annotations report the number of incident pathways. Pathway-level illustration (Right)… view at source ↗
Figure 5
Figure 5. Figure 5: Plate diagram of the BHPI generative model. The generative DAG is: ze → mv,e; Hv,e = zemv,e; {ze, H} → γj,e → µj,e; βj,v = d −1 v P e Hv,eµj,e; yi,v ∼ Bernoulli(σ(αv + x ⊤ i βv)). efficient VI algorithm, we first approximate the posterior by minimizing the Kullback-Leibler (KL) divergence between a variational family q(Θ) and the true posterior, i.e., q ∗ (Θ) = arg min q(Θ)∈Q KL (q(Θ)||p(Θ | Y)), where KL … view at source ↗
Figure 6
Figure 6. Figure 6: Recovery of latent disease hypergraph. Heatmaps show true (left) and inferred (right) disease hypergraph incidence matrices for a representative simulation replicate. BHPI accurately recovers overlapping hypergraph structure with limited redundancy. To qualitatively assess the structural recovery performance of BHPI, we visualize the latent disease hypergraph incidence matrix H and the risk-factor-hyperedg… view at source ↗
Figure 7
Figure 7. Figure 7: Bar plots show true versus inferred effects of predictor 1 across hyperedges, measured by PIP of γ and magnitude µ. B.2. Ablation: Role of the Repulsion Prior [PITH_FULL_IMAGE:figures/full_fig_p018_7.png] view at source ↗
Figure 8
Figure 8. Figure 8: Robustness of predictive performance to repulsion strength across model sizes. Each panel fixes the number of hyperedges E; predictive AUC remains stable across a wide range of λ and E, indicating robustness to both hyperparameters [PITH_FULL_IMAGE:figures/full_fig_p019_8.png] view at source ↗
Figure 9
Figure 9. Figure 9: Structural stability, sharpness, and redundancy across repulsion strengths. B.5. Initialization Stability and Identifiability Empirically, after permutation alignment, the hypergraph overlap across different initializations on the same data is 0.905 ± 0.062 in simulation and 0.875 ± 0.044 on UKB. Cross-replicate overlap in simulation is 0.81 − 0.83. Predictive variability remains small throughout (AUC std … view at source ↗
Figure 10
Figure 10. Figure 10: Robustness to over-specification of the maximum number of hyperedges E, measured by the posterior expected number of active hyperedges [PITH_FULL_IMAGE:figures/full_fig_p020_10.png] view at source ↗
Figure 11
Figure 11. Figure 11: ROC curves of representative diseases for CAVI, logistic regression, and LGBM, with the prevalence indicated in brackets. The Mental Health Pathway (Hyperedge 22): Sleeplessness is a primary driver of a highly specific cluster (V = 1) containing only Other mental disorder. Modulated by Sex and Chest pain, this captures the bidirectional link between sleep and psychiatric instability. (ii) The Cardio-Metab… view at source ↗
read the original abstract

Electronic health records (EHR) pose large-scale multi-disease modeling problems in which many outcomes are rare and strongly influenced by shared risk factors. While modern approaches achieve strong predictive performance, they often treat diseases independently or rely on black-box architectures, offering limited insight into how risk factors organize disease risk and little principled uncertainty quantification. We introduce a Bayesian hypergraph inference framework that reframes multi-disease modeling around latent, risk-factor-modulated disease pathways. Risk factors act on hyperedges, latent disease subsets with shared risk patterns, allowing diseases to participate in multiple distinct pathways and enabling interpretable, higher-order structure beyond pairwise associations. A repulsion prior encourages parsimonious and identifiable structure, while posterior inference provides calibrated uncertainty over both disease groupings and risk-factor influence. To enable scalable inference on large EHR datasets, we develop a structured variational inference algorithm that preserves logical dependencies among hyperedge existence, disease membership, and pathway-level effects. Experiments on simulated data and UK Biobank demonstrate stable and interpretable disease pathway structure, well-calibrated uncertainty, improved estimation for rare diseases, and competitive predictive performance.

Editorial analysis

A structured set of objections, weighed in public.

Desk editor's note, referee report, simulated authors' rebuttal, and a circularity audit. Tearing a paper down is the easy half of reading it; the pith above is the substance, this is the friction.

Referee Report

2 major / 2 minor

Summary. The manuscript introduces a Bayesian hypergraph inference framework for multi-disease modeling from EHR data. Risk factors modulate latent hyperedges representing subsets of diseases with shared risk patterns; diseases may belong to multiple hyperedges. A repulsion prior is used to encourage parsimonious and identifiable structure, while a structured variational inference procedure is developed to perform scalable posterior inference that preserves logical dependencies among hyperedge existence, disease membership, and pathway effects. Experiments on simulated data and UK Biobank are reported to demonstrate interpretable pathway recovery, well-calibrated uncertainty, improved estimation for rare diseases, and competitive predictive performance.

Significance. If the central claims hold, the work would offer a meaningful advance over independent-disease or black-box models by supplying higher-order, interpretable latent structure together with principled uncertainty quantification, which is especially relevant for rare outcomes in large cohorts. The structured variational inference algorithm that preserves logical dependencies among the discrete and continuous components of the model is a clear technical strength.

major comments (2)
  1. [Inference and Experiments sections] The central claim that the repulsion prior together with the structured variational inference produces parsimonious, identifiable hyperedge structure on real EHR data without extensive post-hoc tuning is load-bearing for the interpretability and uncertainty-calibration assertions, yet the manuscript provides no sensitivity analysis on repulsion strength, no explicit identifiability argument, and no diagnostic checks for label-switching or degeneracy when the number of latent pathways is treated as unknown.
  2. [Experiments section] No validation of the variational approximation against exact inference (or against a gold-standard sampler on smaller instances) is reported, and the manuscript does not specify how rare-disease improvements were quantified or whether data splits were pre-specified; these omissions make it impossible to assess whether the reported gains for rare outcomes are robust.
minor comments (2)
  1. [Model section] Notation for hyperedge membership indicators and pathway-level effects should be introduced with an explicit table or diagram early in the model section to improve readability.
  2. The abstract states that the method yields 'stable and interpretable disease pathway structure' on UK Biobank; concrete examples of recovered hyperedges together with their associated risk factors would strengthen the presentation.

Simulated Author's Rebuttal

2 responses · 0 unresolved

We thank the referee for their detailed and constructive comments on our manuscript. We address each of the major comments below and indicate the revisions we will make to strengthen the paper.

read point-by-point responses

  1. Referee: [Inference and Experiments sections] The central claim that the repulsion prior together with the structured variational inference produces parsimonious, identifiable hyperedge structure on real EHR data without extensive post-hoc tuning is load-bearing for the interpretability and uncertainty-calibration assertions, yet the manuscript provides no sensitivity analysis on repulsion strength, no explicit identifiability argument, and no diagnostic checks for label-switching or degeneracy when the number of latent pathways is treated as unknown.

    Authors: We agree that these elements would provide stronger support for our claims. In the revised manuscript, we will add a sensitivity analysis section examining the effect of varying the repulsion prior strength on the inferred number of hyperedges and their stability. We will also include a brief discussion of identifiability, highlighting how the repulsion prior mitigates label-switching by encouraging distinct hyperedge structures. Additionally, we will report diagnostic checks, such as posterior similarity matrices or trace plots for hyperedge assignments, in the supplementary material to assess degeneracy. These changes will be made in the Inference and Experiments sections. revision: yes

  2. Referee: [Experiments section] No validation of the variational approximation against exact inference (or against a gold-standard sampler on smaller instances) is reported, and the manuscript does not specify how rare-disease improvements were quantified or whether data splits were pre-specified; these omissions make it impossible to assess whether the reported gains for rare outcomes are robust.

    Authors: We acknowledge the importance of these validations for assessing the reliability of our results. We will include a new subsection in the Experiments section comparing the structured variational inference to MCMC sampling on smaller simulated datasets where exact inference is feasible. We will also clarify the quantification of rare-disease improvements by specifying the performance metrics (e.g., precision-recall AUC stratified by disease prevalence) and confirm that all data splits were pre-specified prior to analysis, as per the UK Biobank data access protocol described in the methods. These additions will allow readers to better evaluate the robustness of the reported gains. revision: yes

Circularity Check

0 steps flagged

No circularity: generative model with external validation on simulated and UK Biobank data

full rationale

The paper defines a new Bayesian hypergraph model with a repulsion prior and structured variational inference, then evaluates it on simulated data and UK Biobank for pathway recovery, uncertainty calibration, and predictive performance. No equations or claims reduce the reported predictions or identifiability results to quantities fitted on the same target data by construction, nor do any load-bearing steps rely on self-citation chains or imported uniqueness theorems. The framework is presented as generative and externally benchmarked rather than self-referential.

Axiom & Free-Parameter Ledger

2 free parameters · 2 axioms · 1 invented entities

The central claim rests on standard Bayesian modeling assumptions plus several modeling choices introduced for this application. The repulsion prior and the logical-dependency-preserving variational family are domain-specific inventions whose independent support is not shown in the abstract.

free parameters (2)
  • repulsion prior strength
    Hyperparameter controlling sparsity of hyperedges; its value must be chosen or inferred and directly affects the discovered pathway structure.
  • variational family parameters
    Parameters of the structured variational distribution that approximate the posterior over hyperedges and memberships.
axioms (2)
  • domain assumption The data-generating process can be represented as risk factors modulating a collection of latent hyperedges (disease subsets).
    Invoked in the abstract description of the framework; if false, the hypergraph representation does not capture the true risk structure.
  • standard math Standard Bayesian posterior inference is well-defined for the hypergraph model.
    Background assumption required for any Bayesian claim.
invented entities (1)
  • latent hyperedges (risk-factor-modulated disease pathways) no independent evidence
    purpose: To represent higher-order shared-risk groupings that allow diseases to belong to multiple pathways.
    New modeling construct introduced to go beyond pairwise associations; no independent falsifiable handle (e.g., predicted pathway membership testable outside the model) is provided in the abstract.

pith-pipeline@v0.9.1-grok · 5736 in / 1663 out tokens · 21514 ms · 2026-06-27T23:06:54.158909+00:00 · methodology

discussion (0)

Sign in with ORCID, Apple, or X to comment. Anyone can read and Pith papers without signing in.

Reference graph

Works this paper leans on

111 extracted references · 38 canonical work pages

  1. [1]

    The lancet , volume=

    Global burden of 87 risk factors in 204 countries and territories, 1990--2019: a systematic analysis for the Global Burden of Disease Study 2019 , author=. The lancet , volume=. 2020 , publisher=

    1990

  2. [2]

    Bayesian inference for logistic models using P

    Polson, Nicholas G and Scott, James G and Windle, Jesse , journal=. Bayesian inference for logistic models using P. 2013 , publisher=

    2013

  3. [3]

    2006 , publisher=

    Pattern recognition and machine learning , author=. 2006 , publisher=

    2006

  4. [4]

    Journal of Machine Learning Research , author =

    Bayesian. Journal of Machine Learning Research , author =. 2020 , keywords =

    2020

  5. [5]

    Biometrical Journal , author =

    Multivariate frailty models for two types of recurrent events with a dependent terminal event:. Biometrical Journal , author =. 2013 , note =. doi:10.1002/bimj.201200196 , abstract =

    work page doi:10.1002/bimj.201200196 2013

  6. [6]

    Statistical Methods in Medical Research , author =

    Joint analysis of recurrence and termination:. Statistical Methods in Medical Research , author =. 2021 , note =. doi:10.1177/0962280220962522 , abstract =

    work page doi:10.1177/0962280220962522 2021

  7. [7]

    Nature Communications , author =

    Inferring multimodal latent topics from electronic health records , volume =. Nature Communications , author =. 2020 , note =. doi:10.1038/s41467-020-16378-3 , abstract =

    work page doi:10.1038/s41467-020-16378-3 2020

  8. [8]

    Journal of Biomedical Informatics , author =

    Unsupervised machine learning for the discovery of latent disease clusters and patient subgroups using electronic health records , volume =. Journal of Biomedical Informatics , author =. 2020 , keywords =. doi:10.1016/j.jbi.2019.103364 , abstract =

    work page doi:10.1016/j.jbi.2019.103364 2020

  9. [9]

    2024 , keywords =

    Journal of Biomedical Informatics , author =. 2024 , keywords =. doi:10.1016/j.jbi.2024.104638 , abstract =

    work page doi:10.1016/j.jbi.2024.104638 2024

  10. [10]

    Supervised

    Mcauliffe, Jon and Blei, David , year =. Supervised. Advances in

  11. [11]

    Psychological Methods , author =

    Supervised latent. Psychological Methods , author =. 2023 , pages =. doi:10.1037/met0000541 , language =

    work page doi:10.1037/met0000541 2023

  12. [12]

    Marketing Science , author =

    Probabilistic. Marketing Science , author =. 2018 , note =. doi:10.1287/mksc.2018.1113 , abstract =

    work page doi:10.1287/mksc.2018.1113 2018

  13. [13]

    Supervised multi-specialist topic model with applications on large-scale electronic health record data , isbn =

    Song, Ziyang and Toral, Xavier Sumba and Xu, Yixin and Liu, Aihua and Guo, Liming and Powell, Guido and Verma, Aman and Buckeridge, David and Marelli, Ariane and Li, Yue , month = aug, year =. Supervised multi-specialist topic model with applications on large-scale electronic health record data , isbn =. Proceedings of the 12th. doi:10.1145/3459930.346954...

    work page doi:10.1145/3459930.3469543

  14. [14]

    Statistics in medicine , author =

    Bayesian. Statistics in medicine , author =. 2014 , pmid =. doi:10.1002/sim.5964 , abstract =

    work page doi:10.1002/sim.5964 2014

  15. [15]

    Statistical methods in medical research , author =

    Bayesian analysis of multi-type recurrent events and dependent termination with nonparametric covariate functions , volume =. Statistical methods in medical research , author =. 2017 , pmid =. doi:10.1177/0962280215613378 , abstract =

    work page doi:10.1177/0962280215613378 2017

  16. [16]

    Journal of Computational and Graphical Statistics , author =

    A. Journal of Computational and Graphical Statistics , author =. 2020 , note =. doi:10.1080/10618600.2019.1686988 , abstract =

    work page doi:10.1080/10618600.2019.1686988 2020

  17. [17]

    Biometrics , author =

    Accelerated. Biometrics , author =. 2014 , pmid =. doi:10.1111/biom.12163 , abstract =

    work page doi:10.1111/biom.12163 2014

  18. [18]

    Statistics in medicine , author =

    A. Statistics in medicine , author =. 2016 , pmid =. doi:10.1002/sim.7030 , abstract =

    work page doi:10.1002/sim.7030 2016

  19. [19]

    Journal of Data Science , author =

    Bayesian. Journal of Data Science , author =. 2025 , note =. doi:10.6339/25-JDS1182 , abstract =

    work page doi:10.6339/25-jds1182 2025

  20. [20]

    Biostatistics , author =

    Joint modeling of recurrent events and survival: a. Biostatistics , author =. 2020 , keywords =. doi:10.1093/biostatistics/kxy026 , abstract =

    work page doi:10.1093/biostatistics/kxy026 2020

  21. [21]

    doi: 10.1080/01621459.2016.1211016

    Joint. Journal of the American Statistical Association , author =. 2017 , pmid =. doi:10.1080/01621459.2016.1173557 , abstract =

    work page doi:10.1080/01621459.2016.1173557 2017

  22. [22]

    IEEE Transactions on Pattern Analysis and Machine Intelligence , author =

    Advances in. IEEE Transactions on Pattern Analysis and Machine Intelligence , author =. 2019 , keywords =. doi:10.1109/TPAMI.2018.2889774 , abstract =

    work page doi:10.1109/tpami.2018.2889774 2019

  23. [23]

    Advances in Neural Information Processing Systems , volume=

    Amortized variational inference for simple hierarchical models , author=. Advances in Neural Information Processing Systems , volume=

  24. [24]

    Journal of the American Statistical Association , volume=

    Variational inference: A review for statisticians , author=. Journal of the American Statistical Association , volume=. 2017 , publisher=

    2017

  25. [25]

    Proceedings of the 5th International Conference on Learning Representations (ICLR) , year=

    Categorical Reparameterization with Gumbel-Softmax , author=. Proceedings of the 5th International Conference on Learning Representations (ICLR) , year=

  26. [26]

    Journal of the American Statistical Association , volume =

    Variational. Journal of the American Statistical Association , author =. 2022 , note =. doi:10.1080/01621459.2020.1847121 , number =

    work page doi:10.1080/01621459.2020.1847121 2022

  27. [27]

    Bioinformatics , author =

    Variational. Bioinformatics , author =. 2022 , keywords =. doi:10.1093/bioinformatics/btac416 , abstract =

    work page doi:10.1093/bioinformatics/btac416 2022

  28. [28]

    Efficient

    Bai, Jincheng and Song, Qifan and Cheng, Guang , month = nov, year =. Efficient. doi:10.48550/arXiv.2011.07439 , abstract =

    work page doi:10.48550/arxiv.2011.07439 2011

  29. [29]

    Statistics in Medicine , author =

    A shared-parameter continuous-time hidden. Statistics in Medicine , author =. 2019 , note =. doi:10.1002/sim.7994 , abstract =

    work page doi:10.1002/sim.7994 2019

  30. [30]

    Statistics in Medicine , author =

    Applications of continuous time hidden. Statistics in Medicine , author =. 2003 , note =. doi:10.1002/sim.1270 , abstract =

    work page doi:10.1002/sim.1270 2003

  31. [31]

    Additive

    A. Journal of the American Statistical Association , author =. 2020 , note =. doi:10.1080/01621459.2019.1594831 , abstract =

    work page doi:10.1080/01621459.2019.1594831 2020

  32. [32]

    The Annals of Applied Statistics , author =

    Multistate capture–recapture models for irregularly sampled data , volume =. The Annals of Applied Statistics , author =. 2022 , note =. doi:10.1214/21-AOAS1528 , abstract =

    work page doi:10.1214/21-aoas1528 2022

  33. [33]

    Journal of the American Statistical Association , volume =

    Recurrent. Journal of the American Statistical Association , author =. 2021 , pmid =. doi:10.1080/01621459.2020.1801447 , abstract =

    work page doi:10.1080/01621459.2020.1801447 2021

  34. [34]

    The Annals of Applied Statistics , author =

    Bayesian hidden. The Annals of Applied Statistics , author =. 2024 , note =. doi:10.1214/24-AOAS1922 , abstract =

    work page doi:10.1214/24-aoas1922 2024

  35. [35]

    Journal of the Royal Statistical Society Series B: Statistical Methodology , author =

    Semiparametric. Journal of the Royal Statistical Society Series B: Statistical Methodology , author =. 2022 , keywords =. doi:10.1111/rssb.12499 , abstract =

    work page doi:10.1111/rssb.12499 2022

  36. [36]

    Modeling longitudinal dynamics of comorbidities , isbn =

    Maag, Basil and Feuerriegel, Stefan and Kraus, Mathias and Saar-Tsechansky, Maytal and Züger, Thomas , month = apr, year =. Modeling longitudinal dynamics of comorbidities , isbn =. Proceedings of the. doi:10.1145/3450439.3451871 , abstract =

    work page doi:10.1145/3450439.3451871

  37. [37]

    Efficient

    Liu, Yu-Ying and Li, Shuang and Li, Fuxin and Song, Le and Rehg, James M , year =. Efficient. Advances in

  38. [38]

    Biometrics , author =

    Combining. Biometrics , author =. 2023 , keywords =. doi:10.1111/biom.13865 , abstract =

    work page doi:10.1111/biom.13865 2023

  39. [39]

    Journal of Multivariate Analysis , author =

    Continuous time hidden. Journal of Multivariate Analysis , author =. 2020 , keywords =. doi:10.1016/j.jmva.2020.104646 , abstract =

    work page doi:10.1016/j.jmva.2020.104646 2020

  40. [40]

    Statistics in Medicine , author =

    Markov-modulated marked. Statistics in Medicine , author =. 2023 , note =. doi:10.1002/sim.9832 , abstract =

    work page doi:10.1002/sim.9832 2023

  41. [41]

    Biometrics , author =

    Bayesian. Biometrics , author =. 2021 , keywords =. doi:10.1111/biom.13261 , abstract =

    work page doi:10.1111/biom.13261 2021

  42. [42]

    Biometrics , author =

    A joint model for multistate disease processes and random informative observation times, with applications to electronic medical records data , volume =. Biometrics , author =. 2015 , pmid =. doi:10.1111/biom.12252 , abstract =

    work page doi:10.1111/biom.12252 2015

  43. [43]

    doi:10.24963/ijcai.2021/353 , abstract =

    Huang, Jing and Yang, Jie , month = aug, year =. doi:10.24963/ijcai.2021/353 , abstract =

    work page doi:10.24963/ijcai.2021/353 2021

  44. [44]

    Journal of the American Statistical Association , year =

    Cohesion and. Journal of the American Statistical Association , author =. 2024 , note =. doi:10.1080/01621459.2023.2191821 , abstract =

    work page doi:10.1080/01621459.2023.2191821 2024

  45. [45]

    From fixed-x to random-x regression: Bias-variance decompositions, covariance penalties, and prediction error estimation,

    Bayesian. Journal of the American Statistical Association , author =. 2020 , note =. doi:10.1080/01621459.2018.1537918 , abstract =

    work page doi:10.1080/01621459.2018.1537918 2020

  46. [46]

    Journal of the Royal Statistical Society Series B: Statistical Methodology , author =

    Repulsion, chaos, and equilibrium in mixture models , volume =. Journal of the Royal Statistical Society Series B: Statistical Methodology , author =. 2025 , keywords =. doi:10.1093/jrsssb/qkae096 , abstract =

    work page doi:10.1093/jrsssb/qkae096 2025

  47. [47]

    Advances in Data Analysis and Classification , author =

    Model-based clustering for random hypergraphs , volume =. Advances in Data Analysis and Classification , author =. 2022 , keywords =. doi:10.1007/s11634-021-00454-7 , abstract =

    work page doi:10.1007/s11634-021-00454-7 2022

  48. [48]

    The Lancet , volume=

    Epidemiology of multimorbidity and implications for health care, research, and medical education: a cross-sectional study , author=. The Lancet , volume=. 2012 , publisher=

    2012

  49. [49]

    Proceedings of the National Academy of Sciences , volume=

    The human disease network , author=. Proceedings of the National Academy of Sciences , volume=. 2007 , publisher=

    2007

  50. [50]

    Nature Reviews Genetics , volume=

    Network medicine: a network-based approach to human disease , author=. Nature Reviews Genetics , volume=. 2011 , publisher=

    2011

  51. [51]

    A Dynamic Network Approach for the Study of Human , volume =

    Blumm, Nicholas and Barabasi, Albert-Laszlo and Christakis, Nicholas , year =. A Dynamic Network Approach for the Study of Human , volume =. PLoS computational biology , doi =

  52. [52]

    Science , volume=

    Uncovering disease-disease relationships through the incomplete interactome , author=. Science , volume=. 2015 , publisher=

    2015

  53. [53]

    Scientific reports , volume=

    Deep patient: an unsupervised representation to predict the future of patients from the electronic health records , author=. Scientific reports , volume=. 2016 , publisher=

    2016

  54. [54]

    Machine learning for healthcare conference , pages=

    Doctor ai: Predicting clinical events via recurrent neural networks , author=. Machine learning for healthcare conference , pages=. 2016 , organization=

    2016

  55. [55]

    NPJ digital medicine , volume=

    Scalable and accurate deep learning with electronic health records , author=. NPJ digital medicine , volume=. 2018 , publisher=

    2018

  56. [56]

    Technometrics , volume=

    Ridge regression: Biased estimation for nonorthogonal problems , author=. Technometrics , volume=. 1970 , publisher=

    1970

  57. [57]

    Journal of the Royal Statistical Society Series B: Statistical Methodology , volume=

    Regression shrinkage and selection via the lasso , author=. Journal of the Royal Statistical Society Series B: Statistical Methodology , volume=. 1996 , publisher=

    1996

  58. [58]

    Journal of the Royal Statistical Society: Series B (Statistical Methodology) , volume=

    Regularization and variable selection via the elastic net , author=. Journal of the Royal Statistical Society: Series B (Statistical Methodology) , volume=. 2005 , publisher=

    2005

  59. [59]

    Political Analysis , year=

    Logistic Regression in Rare Events Data , author=. Political Analysis , year=

  60. [60]

    Biometrika , volume=

    Bias reduction of maximum likelihood estimates , author=. Biometrika , volume=. 1993 , publisher=

    1993

  61. [61]

    Statistics in medicine , volume=

    A solution to the problem of separation in logistic regression , author=. Statistics in medicine , volume=. 2002 , publisher=

    2002

  62. [62]

    A weakly informative default prior distribution for logistic and other regression models , volume =

    Gelman, Andrew and Jakulin, Aleks and Pittau, Maria Grazia and Su, Yu-Sung , year =. A weakly informative default prior distribution for logistic and other regression models , volume =. The Annals of Applied Statistics , doi =

  63. [63]

    Machine Learning , volume=

    Multitask learning , author=. Machine Learning , volume=. 1997 , publisher=

    1997

  64. [64]

    Proceedings of the tenth ACM SIGKDD international conference on Knowledge discovery and data mining , pages=

    Regularized multi--task learning , author=. Proceedings of the tenth ACM SIGKDD international conference on Knowledge discovery and data mining , pages=

  65. [65]

    Machine Learning , volume=

    Convex multi-task feature learning , author=. Machine Learning , volume=. 2008 , publisher=

    2008

  66. [66]

    Advances in neural information processing systems , volume=

    Multi-task Gaussian process prediction , author=. Advances in neural information processing systems , volume=

  67. [67]

    Scientific data , volume=

    Multitask learning and benchmarking with clinical time series data , author=. Scientific data , volume=. 2019 , publisher=

    2019

  68. [68]

    International Journal of Data Warehousing and Mining , volume=

    Multi-label classification: An overview , author=. International Journal of Data Warehousing and Mining , volume=. 2007 , publisher=

    2007

  69. [69]

    Machine Learning , volume=

    Classifier chains for multi-label classification , author=. Machine Learning , volume=. 2011 , publisher=

    2011

  70. [70]

    IEEE Transactions on Knowledge and Data Engineering , volume=

    A review on multi-label learning algorithms , author=. IEEE Transactions on Knowledge and Data Engineering , volume=. 2014 , publisher=

    2014

  71. [71]

    Journal of Machine Learning Research , volume=

    Latent Dirichlet allocation , author=. Journal of Machine Learning Research , volume=

  72. [72]

    Advances in neural information processing systems , volume=

    Correlated topic models , author=. Advances in neural information processing systems , volume=. 2006 , publisher=

    2006

  73. [73]

    Pachinko allocation:

    Li, Wei and McCallum, Andrew , booktitle=. Pachinko allocation:

  74. [74]

    Advances in neural information processing systems , volume=

    Learning with hypergraphs: Clustering, classification, and embedding , author=. Advances in neural information processing systems , volume=

  75. [75]

    Proceedings of the AAAI conference on artificial intelligence , volume=

    Hypergraph neural networks , author=. Proceedings of the AAAI conference on artificial intelligence , volume=

  76. [76]

    Advances in neural information processing systems , volume=

    Hypergcn: A new method for training graph convolutional networks on hypergraphs , author=. Advances in neural information processing systems , volume=

  77. [77]

    Hypergraph Transformers for

    Xu, Ran and Ali, Mohammed K and Ho, Joyce C and Yang, Carl , journal=. Hypergraph Transformers for

  78. [78]

    Bycroft, Clare and Freeman, Colin and Petkova, Desislava and Band, Gavin and Elliott, Lloyd T and Sharp, Kevin and Motyer, Allan and Vukcevic, Damjan and Delaneau, Olivier and O’Connell, Jared and others , journal=. The. 2018 , publisher=

    2018

  79. [79]

    Nature Reviews Disease Primers , volume=

    Multimorbidity , author=. Nature Reviews Disease Primers , volume=. 2022 , doi=

    2022

  80. [80]

    The Lancet , volume=

    Global burden of 87 risk factors in 204 countries and territories, 1990--2019: a systematic analysis for the Global Burden of Disease Study 2019 , author=. The Lancet , volume=. 2020 , publisher=

    1990

Showing first 80 references.