pith. sign in

arxiv: 2605.16927 · v1 · pith:L6REGX6Onew · submitted 2026-05-16 · 💻 cs.AI

From Static Risk to Dynamic Trajectories: Toward World-Model-Inspired Clinical Prediction

Pujun Feng , Xiaoyu Guo , Seyed Ehsan Saffari , Min Hun Lee , Siew-Kei Lam , Erik Cambria , Xibin Sun , Yangtao Zhou
show 5 more authors
Tong Yang Xiaoyu Zhang Tao Tan Yue Sun Bin Cui
This is my paper

Pith reviewed 2026-05-19 20:34 UTC · model grok-4.3

classification 💻 cs.AI
keywords clinical predictiondisease trajectoriescounterfactual estimationpolicy evaluationtime-varying confoundinglongitudinal causal inferenceobservation biasintervention-aware modeling
0
0 comments X

The pith

A unified framework links patient forecasts, counterfactual treatment paths, and policy checks by jointly modeling disease, treatment choices, and observation biases.

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

The paper organizes clinical prediction methods around six linked parts: three decision tasks of factual forecasting, counterfactual estimation, and policy evaluation, plus three data-generating mechanisms of disease evolution, treatment assignment, and the observation process. This structure creates a single framework that works across discrete and continuous time while directly handling treatment assignment, time-varying confounding, and observation bias. A sympathetic reader would care because static risk models trained on routine care data mix disease biology with clinician behavior, producing estimates that break down once treatments change or observations become irregular. The approach moves the field from isolated prediction toward decision-grade evidence that supports individualized treatment simulations and policy testing before deployment.

Core claim

The paper presents the first unified framework bridging factual forecasting, counterfactual trajectory estimation, and policy evaluation across discrete and continuous time, explicitly addressing treatment assignment, time-varying confounding, and observation bias. It organizes the literature by mapping method families such as multistate and joint models, temporal point processes, deep sequence architectures, and longitudinal causal inference onto the six components that determine identifiability.

What carries the argument

The six linked components—three decision tasks (factual forecasting, counterfactual estimation, policy evaluation) and three data-generating mechanisms (disease evolution, treatment assignment, observation process)—that together determine identifiability and organize existing methods.

If this is right

  • Treatment-sensitive individualized future trajectories become estimable from observational records.
  • Alternative treatment policies can be stress-tested before real-world deployment.
  • Closed-loop learning health systems gain criteria for when to adapt or abstain based on evidence strength.
  • Evaluation can align with claim strength using overlap diagnostics, uncertainty measures, and target-trial emulation.

Where Pith is reading between the lines

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

  • The same structure could guide integration of high-frequency sensor data into trajectory models without breaking existing identifiability results.
  • It suggests a route to embed these components inside reinforcement-learning agents that learn safe intervention policies from electronic health records.

Load-bearing premise

That these six components are enough to determine identifiability and to map all prior method families without losing important structure or coverage.

What would settle it

A real clinical dataset or prediction task in which a known source of bias or an established method cannot be placed inside any of the six components without material loss of accuracy or coverage.

Figures

Figures reproduced from arXiv: 2605.16927 by Bin Cui, Erik Cambria, Min Hun Lee, Pujun Feng, Seyed Ehsan Saffari, Siew-Kei Lam, Tao Tan, Tong Yang, Xiaoyu Guo, Xiaoyu Zhang, Xibin Sun, Yangtao Zhou, Yue Sun.

Figure 1
Figure 1. Figure 1: World-model-based prediction and decision-making in clinical trajectories. Multimodal clinical data are [PITH_FULL_IMAGE:figures/full_fig_p002_1.png] view at source ↗
Figure 2
Figure 2. Figure 2: Evolution of disease trajectory modeling in medicine and AI. [PITH_FULL_IMAGE:figures/full_fig_p003_2.png] view at source ↗
Figure 3
Figure 3. Figure 3: Overview of our intervention-aware disease trajectory modeling framework. [PITH_FULL_IMAGE:figures/full_fig_p005_3.png] view at source ↗
Figure 4
Figure 4. Figure 4: PRISMA diagram for the screening process. [PITH_FULL_IMAGE:figures/full_fig_p006_4.png] view at source ↗
Figure 5
Figure 5. Figure 5: Overview: from the natural history of disease to data-driven trajectories. [PITH_FULL_IMAGE:figures/full_fig_p007_5.png] view at source ↗
Figure 6
Figure 6. Figure 6: Discussion overview: assumptions, failure modes, and guardrails for intervention-aware disease trajectory [PITH_FULL_IMAGE:figures/full_fig_p021_6.png] view at source ↗
read the original abstract

Clinical decision-making is a feedback system where risk estimates influence treatment, which in turn changes disease trajectories, and both shape clinicians' measurement practices. Static prediction often fails clinically: models trained on observational care logs conflate disease biology with clinician behavior, particularly under treatment confounder feedback and irregular or informative observation. This Review focuses on intervention-aware disease trajectory modeling in clinical AI--methods estimating patient-specific longitudinal disease evolution and assessing trajectory changes under alternative treatments. We organize the field around six linked components: three decision tasks (factual forecasting, counterfactual estimation, policy evaluation) and three data-generating mechanisms (disease evolution, treatment assignment, observation process) that determine identifiability. We present the first unified framework bridging forecasting, counterfactual trajectories, and policy evaluation across discrete/continuous time, explicitly addressing treatment assignment, time-varying confounding, and observation bias. We synthesize key method families (multistate/joint models, temporal point-process, deep sequence architectures, longitudinal causal inference), map them to relevant components, and align evaluation with claim strength via overlap diagnostics, uncertainty quantification, off-policy robustness, and target-trial validation. This synthesis advances benchmark prediction to decision-grade clinical evidence, enabling treatment-sensitive individualized futures, pre-deployment policy stress-testing, and safer closed-loop learning health systems that adapt/abstain when evidence is insufficient.

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 paper is a review proposing a unified framework for intervention-aware disease trajectory modeling in clinical AI. It organizes the literature around six linked components—three decision tasks (factual forecasting, counterfactual estimation, policy evaluation) and three data-generating mechanisms (disease evolution, treatment assignment, observation process)—that are said to determine identifiability. The framework is presented as bridging forecasting, counterfactual trajectories, and policy evaluation across discrete and continuous time while addressing treatment assignment, time-varying confounding, and observation bias. It synthesizes method families including multistate/joint models, temporal point processes, deep sequence architectures, and longitudinal causal inference, and aligns evaluation strategies (overlap diagnostics, uncertainty quantification, off-policy robustness, target-trial validation) with claim strength.

Significance. If the six-component taxonomy provides a comprehensive and non-redundant mapping of existing methods without material omissions and supplies a sound basis for identifiability, the synthesis would be a useful organizational contribution. It could help shift clinical prediction from static risk scores toward dynamic, treatment-sensitive trajectory modeling and support safer policy evaluation in learning health systems. The explicit linkage of decision tasks to data-generating mechanisms and the emphasis on evaluation diagnostics matched to claim strength are constructive elements.

major comments (2)
  1. [Abstract and framework description] Abstract and framework description: the central claim that the six linked components suffice to determine identifiability across discrete/continuous time and close all relevant identification gaps (including time-varying confounding and observation bias) is not supported by any explicit identifiability result, such as a g-computation formula or potential-outcome theorem stated under the joint disease-evolution, treatment-assignment, and observation mechanisms. This is load-bearing for the assertion of a unified framework that comprehensively maps method families without loss of structure.
  2. [Framework section] Framework section: the weakest assumption—that the six components are jointly sufficient to map all relevant prior literature without material omissions—is not tested against structures such as dependent censoring, continuous-time positivity violations, or informative measurement processes that may fall outside the proposed taxonomy. A concrete check against at least one such structure would be required to substantiate the completeness claim.
minor comments (2)
  1. [Abstract] Abstract: the phrasing 'first unified framework' should be qualified by a brief comparison to prior syntheses in longitudinal causal inference to avoid overstatement.
  2. [Evaluation discussion] Evaluation discussion: the alignment of diagnostics (overlap, uncertainty, off-policy robustness, target-trial validation) with claim strength is described at a high level; a table mapping specific diagnostics to each of the three decision tasks would improve clarity.

Simulated Author's Rebuttal

2 responses · 0 unresolved

We thank the referee for the thoughtful and constructive review. The comments correctly identify areas where the presentation of identifiability and completeness claims can be strengthened. We respond to each major comment below and commit to revisions that address the concerns without altering the core synthesis.

read point-by-point responses

  1. Referee: [Abstract and framework description] Abstract and framework description: the central claim that the six linked components suffice to determine identifiability across discrete/continuous time and close all relevant identification gaps (including time-varying confounding and observation bias) is not supported by any explicit identifiability result, such as a g-computation formula or potential-outcome theorem stated under the joint disease-evolution, treatment-assignment, and observation mechanisms. This is load-bearing for the assertion of a unified framework that comprehensively maps method families without loss of structure.

    Authors: We acknowledge that the manuscript does not derive a new identifiability theorem. As a review, it synthesizes and links existing results from the causal inference literature to the three data-generating mechanisms. In the revision we will add a concise subsection in the Framework section that explicitly states the relevant identifiability conditions for each decision task, referencing standard results such as the longitudinal g-computation formula under time-varying confounding and the assumptions required for identification when disease evolution, treatment assignment, and observation processes are jointly modeled. This will make the linkage to established theorems explicit while preserving the review character of the work. revision: yes

  2. Referee: [Framework section] Framework section: the weakest assumption—that the six components are jointly sufficient to map all relevant prior literature without material omissions—is not tested against structures such as dependent censoring, continuous-time positivity violations, or informative measurement processes that may fall outside the proposed taxonomy. A concrete check against at least one such structure would be required to substantiate the completeness claim.

    Authors: We agree that a concrete mapping exercise would strengthen the completeness claim. In the revised Framework section we will include a short discussion that maps dependent censoring and informative measurement processes onto the observation-process component, noting the additional conditional-independence assumptions typically required. We will also indicate how continuous-time positivity violations are diagnosed within the overlap-diagnostics component already described. This provides an explicit check against the cited structures without expanding the taxonomy. revision: yes

Circularity Check

0 steps flagged

Conceptual synthesis of prior methods without self-referential derivations

full rationale

The paper is a review synthesizing existing literature on intervention-aware clinical prediction. It organizes methods around six linked components (three decision tasks and three data-generating mechanisms) that it states determine identifiability, but provides no new equations, fitted parameters, or first-principles derivations that reduce to its own inputs by construction. The framework bridges forecasting, counterfactuals, and policy evaluation by mapping prior families (multistate models, point processes, deep architectures, causal inference) to these components, relying on external benchmarks rather than internal self-definition or self-citation chains. No uniqueness theorems, ansatzes, or renamings of known results are introduced as load-bearing steps. The analysis is therefore self-contained against external literature with no circular reductions.

Axiom & Free-Parameter Ledger

0 free parameters · 1 axioms · 0 invented entities

The paper is a conceptual review that introduces no new free parameters or invented entities. It relies on standard domain assumptions from causal inference and longitudinal modeling.

axioms (1)
  • domain assumption The six linked components (three decision tasks and three data-generating mechanisms) determine identifiability for intervention-aware modeling.
    Invoked when organizing the field and claiming the unified framework bridges forecasting, counterfactuals, and policy evaluation.

pith-pipeline@v0.9.0 · 5805 in / 1269 out tokens · 67131 ms · 2026-05-19T20:34:32.107867+00:00 · methodology

discussion (0)

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

Lean theorems connected to this paper

Citations machine-checked in the Pith Canon. Every link opens the source theorem in the public Lean library.

What do these tags mean?
matches
The paper's claim is directly supported by a theorem in the formal canon.
supports
The theorem supports part of the paper's argument, but the paper may add assumptions or extra steps.
extends
The paper goes beyond the formal theorem; the theorem is a base layer rather than the whole result.
uses
The paper appears to rely on the theorem as machinery.
contradicts
The paper's claim conflicts with a theorem or certificate in the canon.
unclear
Pith found a possible connection, but the passage is too broad, indirect, or ambiguous to say the theorem truly supports the claim.

Reference graph

Works this paper leans on

107 extracted references · 107 canonical work pages · 3 internal anchors

  1. [1]

    An empirical transition matrix for non-homogeneous markov chains based on censored observations.Scandinavian Journal of Statistics, 5(3):141–150, 1978

    Odd Aalen and Søren Johansen. An empirical transition matrix for non-homogeneous markov chains based on censored observations.Scandinavian Journal of Statistics, 5(3):141–150, 1978

    work page 1978

  2. [2]

    Analyzing patient trajectories with artificial intelligence.Journal of medical internet research, 23(12):e29812, 2021

    Ahmed Allam, Stefan Feuerriegel, Michael Rebhan, and Michael Krauthammer. Analyzing patient trajectories with artificial intelligence.Journal of medical internet research, 23(12):e29812, 2021

    work page 2021

  3. [3]

    Gill, and Niels Keiding.Statistical Models Based on Counting Processes

    Per Kragh Andersen, Ørnulf Borgan, Richard D. Gill, and Niels Keiding.Statistical Models Based on Counting Processes. Springer, New York, 1993

    work page 1993

  4. [4]

    Angrist, Guido W

    Joshua D. Angrist, Guido W. Imbens, and Donald B. Rubin. Identification of causal effects using instrumental variables.Journal of the American Statistical Association, 91(434):444–455, 1996

    work page 1996

  5. [5]

    Austin, Frank E

    Peter C. Austin, Frank E. Harrell, and Davide van Klaveren. Graphical calibration curves and the integrated calibration index (ICI) for survival models.Statistics in Medicine, 39(21):2714–2742, 2020

    work page 2020

  6. [6]

    E. Badidi. Edge ai for early detection of chronic diseases and the spread of infectious diseases: Opportunities, challenges, and future directions.Future Internet, 15(11):370, 2023

    work page 2023

  7. [7]

    S. Bai, J. Z. Kolter, and V . Koltun. An empirical evaluation of generic convolutional and recurrent networks for sequence modeling.arXiv preprint arXiv:1803.01271, 2018. 25 arXivTemplate

    work page internal anchor Pith review Pith/arXiv arXiv 2018

  8. [8]

    Mike Baiocchi, Jing Cheng, and Dylan S. Small. Instrumental variable methods for causal inference.Statistics in Medicine, 33(13):2297–2340, 2014

    work page 2014

  9. [9]

    Patient subtyping via time-aware LSTM networks

    Inci M Baytas, Cao Xiao, Xi Zhang, Fei Wang, Anil K Jain, and Jiayu Zhou. Patient subtyping via time-aware LSTM networks. InProceedings of the 23rd ACM SIGKDD International Conference on Knowledge Discovery and Data Mining, pages 65–74, 2017

    work page 2017

  10. [10]

    Estimating counterfactual treatment outcomes over time through adversarially balanced representations.arXiv preprint arXiv:2002.04083, 2020

    Ioana Bica, Ahmed M Alaa, James Jordon, and Mihaela Van Der Schaar. Estimating counterfactual treatment outcomes over time through adversarially balanced representations.arXiv preprint arXiv:2002.04083, 2020

    work page arXiv 2002

  11. [11]

    Alaa, James Jordon, and Mihaela van der Schaar

    Ioana Bica, Ahmed M. Alaa, James Jordon, and Mihaela van der Schaar. Time series deconfounder: Estimating treatment effects over time in the presence of hidden confounders. In Hal Daumé III and Aarti Singh, editors, Proceedings of the 37th International Conference on Machine Learning, volume 119 ofProceedings of Machine Learning Research, pages 884–895. P...

    work page 2020

  12. [12]

    Ioana Bica, Ahmed M Alaa, Craig Lambert, and Mihaela Van Der Schaar. From real-world patient data to individualized treatment effects using machine learning: current and future methods to address underlying challenges.Clinical Pharmacology & Therapeutics, 109(1):87–100, 2021

    work page 2021

  13. [13]

    Learning the natural history of human disease with generative transformers.Nature, 644(8071):480–484, 2025

    Ioana Bica, Alexander Ulrich, Feiran Huang, Kasper Søholm, Steffen Bülow, Anders Bertelsen, et al. Learning the natural history of human disease with generative transformers.Nature, 644(8071):480–484, 2025

    work page 2025

  14. [14]

    Causal dynamic variational autoencoder for counterfactual regression in longitudinal data, 2025

    Mouad El Bouchattaoui, Myriam Tami, Benoit Lepetit, and Paul-Henry Cournède. Causal dynamic variational autoencoder for counterfactual regression in longitudinal data, 2025

    work page 2025

  15. [15]

    Estimating treatment effects in continuous time with hidden confounders, 2023

    Defu Cao, James Enouen, and Yan Liu. Estimating treatment effects in continuous time with hidden confounders, 2023

    work page 2023

  16. [16]

    Estimating treatment effects in the presence of irregular time series observations with hidden confounders

    Defu Cao, James Enouen, Yujing Wang, Xiangchen Song, Chuizheng Meng, Hao Niu, and Yan Liu. Estimating treatment effects in the presence of irregular time series observations with hidden confounders. InProceedings of the 37th AAAI Conference on Artificial Intelligence, pages 6897–6905, 2023

    work page 2023

  17. [17]

    Carrasco-Ribelles, José Llanes-Jurado, Carlos Gallego-Moll, Margarita Cabrera-Bean, Mònica Monteagudo-Zaragoza, Concepción Violán, and Edurne Zabaleta-del Olmo

    Lucía A. Carrasco-Ribelles, José Llanes-Jurado, Carlos Gallego-Moll, Margarita Cabrera-Bean, Mònica Monteagudo-Zaragoza, Concepción Violán, and Edurne Zabaleta-del Olmo. Prediction models using arti- ficial intelligence and longitudinal data from electronic health records: a systematic methodological review. Journal of the American Medical Informatics Ass...

    work page 2072

  18. [18]

    Caterini and Dong Eui Chang.Recurrent Neural Networks, pages 59–79

    Anthony L. Caterini and Dong Eui Chang.Recurrent Neural Networks, pages 59–79. Springer International Publishing, Cham, 2018

    work page 2018

  19. [19]

    Recurrent neural networks for multivariate time series with missing values.Scientific Reports, 8(1):6085, 2018

    Zhengping Che, Sanjay Purushotham, Kyunghyun Cho, David Sontag, and Yan Liu. Recurrent neural networks for multivariate time series with missing values.Scientific Reports, 8(1):6085, 2018

    work page 2018

  20. [20]

    Deep learning with multimodal representation for pancancer prognosis prediction.Nature Communications, 10(1):4008, 2019

    Abhigyan Cheerla and Olivier Gevaert. Deep learning with multimodal representation for pancancer prognosis prediction.Nature Communications, 10(1):4008, 2019

    work page 2019

  21. [21]

    Ricky T. Q. Chen, Yulia Rubanova, Jesse Bettencourt, and David K. Duvenaud. Neural ordinary differential equations, 2018

    work page 2018

  22. [22]

    Integrative analysis of histopathological images and genomic data predicts clear cell renal cell carcinoma prognosis.Cancer Research, 77(21):e91–e100, 2017

    Jun Cheng, Jianfei Zhang, Yaozong Han, Xin Wang, Xiaonan Ye, Yan Meng, Anil Parwani, Kun Huang, Liangliang Cao, Dong Ni, and et al. Integrative analysis of histopathological images and genomic data predicts clear cell renal cell carcinoma prognosis.Cancer Research, 77(21):e91–e100, 2017

    work page 2017

  23. [23]

    K. Cho, B. van Merriënboer, C. Gulcehre, et al. Learning phrase representations using rnn encoder-decoder for statistical machine translation.arXiv preprint arXiv:1406.1078, 2014

    work page internal anchor Pith review Pith/arXiv arXiv 2014

  24. [24]

    Stewart, and Jimeng Sun

    Edward Choi, Mohammad Taha Bahadori, Andy Schuetz, Walter F. Stewart, and Jimeng Sun. Doctor ai: Predicting clinical events via recurrent neural networks. InProceedings of the 1st Machine Learning for Healthcare Conference, volume 56 ofProceedings of Machine Learning Research, pages 301–318. PMLR, 2016

    work page 2016

  25. [25]

    Edward Choi, Mohammad Taha Bahadori, Jimeng Sun, Joshua Kulas, Andy Schuetz, and Walter F. Stewart. RETAIN: An interpretable predictive model for healthcare using reverse time attention mechanism. InAdvances in Neural Information Processing Systems, 2016

    work page 2016

  26. [26]

    GRAM: Graph-based attention model for healthcare representation learning

    Edward Choi, Cao Xiao, Walter Stewart, and Jimeng Sun. GRAM: Graph-based attention model for healthcare representation learning. InProceedings of the 23rd ACM SIGKDD International Conference on Knowledge Discovery and Data Mining, pages 787–795, 2017

    work page 2017

  27. [27]

    P. M. D. S. Costa, S. L. A. Sales, D. P. Pinheiro, et al. Epigenetic reprogramming in cancer: From diagnosis to treatment.Frontiers in Cell and Developmental Biology, 11:1116805, 2023. 26 arXivTemplate

    work page 2023

  28. [28]

    Using natural experiments to evaluate population health interventions: new MRC guidance.Journal of Epidemiology and Community Health, 66(12):1182–1186, 2012

    Peter Craig, Chris Cooper, David Gunnell, Sally Haw, Kenny Lawson, Sally Macintyre, David Ogilvie, Mark Petticrew, Aaron Reeves, Matt Sutton, and Steve Thompson. Using natural experiments to evaluate population health interventions: new MRC guidance.Journal of Epidemiology and Community Health, 66(12):1182–1186, 2012

    work page 2012

  29. [29]

    Predicting the impact of treatments over time with uncertainty aware neural differential equations

    Edward de Brouwer, Javier Gonzalez, and Stephanie Hyland. Predicting the impact of treatments over time with uncertainty aware neural differential equations. In Gustau Camps-Valls, Francisco J. R. Ruiz, and Isabel Valera, editors,Proceedings of the 25th International Conference on Artificial Intelligence and Statistics, volume 151 of Proceedings of Machin...

    work page 1917

  30. [30]

    Recurrent marked temporal point processes: Embedding event history to vector

    Nan Du, Yichen Wang, Niao He, and Le Song. Recurrent marked temporal point processes: Embedding event history to vector. InProceedings of the 22nd ACM SIGKDD International Conference on Knowledge Discovery and Data Mining, pages 1555–1564, New York, NY , USA, 2016. Association for Computing Machinery

    work page 2016

  31. [31]

    Causal contrastive learning for counterfactual regression over time.Advances in Neural Information Processing Systems, 37:1333–1369, 2024

    Mouad El Bouchattaoui, Myriam Tami, Benoit Lepetit, and Paul-Henry Cournède. Causal contrastive learning for counterfactual regression over time.Advances in Neural Information Processing Systems, 37:1333–1369, 2024

    work page 2024

  32. [32]

    Eskofier and Jochen Klucken

    Bjoern M. Eskofier and Jochen Klucken. Predictive models for health deterioration: Understanding disease pathways for personalized medicine.Annual Review of Biomedical Engineering, 25:131–156, 2023

    work page 2023

  33. [33]

    Coopersmith, Craig French, Flavia R

    Laura Evans, Andrew Rhodes, Waleed Alhazzani, Massimo Antonelli, Craig M. Coopersmith, Craig French, Flavia R. Machado, Lauralyn McIntyre, Marlies Ostermann, Hallie C. Prescott, Christa Schorr, Stephen Simpson, W. Joost Wiersinga, Fayez Alshamsi, Derek C. Angus, Yaseen Arabi, Luciano Azevedo, Richard Beale, Greg Beilman, Emilie Belley-Cote, Lisa Burry, Ma...

    work page 2021

  34. [34]

    F. P. Fabrizio. Unlocking lung cancer cell dormancy: An epigenetic perspective.International Journal of Molecular Sciences, 26(22):10997, 2025

    work page 2025

  35. [35]

    Exploring the role of ai in predicting chronic disease progression: diabetes and cardiovascular diseases.Premier Journal of Public Health, 4:100021, 2025

    Azza Moustafa Fahmy. Exploring the role of ai in predicting chronic disease progression: diabetes and cardiovascular diseases.Premier Journal of Public Health, 4:100021, 2025

    work page 2025

  36. [36]

    Causal machine learning for predicting treatment outcomes.Nature Medicine, 2024

    Stefan Feuerriegel et al. Causal machine learning for predicting treatment outcomes.Nature Medicine, 2024

    work page 2024

  37. [37]

    Toward a science of learning systems: a research agenda for the high-functioning learning health system.Journal of the American Medical Informatics Association, 22(1):43–50, 2015

    Charles Friedman, James Rubin, Julie Brown, Melinda Buntin, Michael Corn, Lynn Etheredge, Cynthia Gunter, Mark Musen, Richard Platt, William Stead, Kevin Sullivan, and Douglas Van Houweling. Toward a science of learning systems: a research agenda for the high-functioning learning health system.Journal of the American Medical Informatics Association, 22(1)...

    work page 2015

  38. [38]

    Gerds and Martin Schumacher

    Thomas A. Gerds and Martin Schumacher. Consistent estimation of the expected Brier score in general survival models with right-censored event times.Biometrical Journal, 48(6):1029–1040, 2006

    work page 2006

  39. [39]

    Guidelines for reinforcement learning in healthcare.Nature Medicine, 25(1):16–18, 2019

    Omer Gottesman, Fredrik Johansson, Matthieu Komorowski, Aldo Faisal, David Sontag, Finale Doshi-Velez, and Leo Anthony Celi. Guidelines for reinforcement learning in healthcare.Nature Medicine, 25(1):16–18, 2019

    work page 2019

  40. [40]

    Assessment and comparison of prognostic classification schemes for survival data.Statistics in Medicine, 18(17-18):2529–2545, 1999

    Erika Graf, Claudia Schmoor, Willi Sauerbrei, and Martin Schumacher. Assessment and comparison of prognostic classification schemes for survival data.Statistics in Medicine, 18(17-18):2529–2545, 1999

    work page 1999

  41. [41]

    Graves.Supervised Sequence Labelling with Recurrent Neural Networks

    A. Graves.Supervised Sequence Labelling with Recurrent Neural Networks. 2012

    work page 2012

  42. [42]

    A multi-center study on the adaptability of a shared foundation model for electronic health records.NPJ Digital Medicine, 7(1):171, 2024

    Lin Lawrence Guo, Jason Fries, Ethan Steinberg, Scott Lanyon Fleming, Keith Morse, Catherine Aftandilian, Jose Posada, Nigam Shah, and Lillian Sung. A multi-center study on the adaptability of a shared foundation model for electronic health records.NPJ Digital Medicine, 7(1):171, 2024

    work page 2024

  43. [43]

    Estimating the treatment effect over time under general interference through deep learner integrated tmle, 2024

    Suhan Guo, Furao Shen, and Ni Li. Estimating the treatment effect over time under general interference through deep learner integrated tmle, 2024

    work page 2024

  44. [44]

    Alan G. Hawkes. Spectra of some self-exciting and mutually exciting point processes.Biometrika, 58(1):83–90, 1971

    work page 1971

  45. [45]

    Heagerty, Thomas Lumley, and Margaret S

    Patrick J. Heagerty, Thomas Lumley, and Margaret S. Pepe. Time-dependent roc curves for censored survival data and a diagnostic marker.Biometrics, 56(2):337–344, 2000. 27 arXivTemplate

    work page 2000

  46. [46]

    Heagerty and Yingye Zheng

    Patrick J. Heagerty and Yingye Zheng. Survival model predictive accuracy and roc curves.Biometrics, 61(1):92– 105, 2005

    work page 2005

  47. [47]

    Using big data to emulate a target trial when a randomized trial is not available.American Journal of Epidemiology, 2016

    Miguel A Hernán and James M Robins. Using big data to emulate a target trial when a randomized trial is not available.American Journal of Epidemiology, 2016

    work page 2016

  48. [48]

    Chapman & Hall/CRC, Boca Raton, 2020

    Miguel A Hernán and James M Robins.Causal Inference: What If. Chapman & Hall/CRC, Boca Raton, 2020

    work page 2020

  49. [49]

    Stabilized neural prediction of potential outcomes in continuous time, 2024

    Konstantin Hess and Stefan Feuerriegel. Stabilized neural prediction of potential outcomes in continuous time, 2024

    work page 2024

  50. [50]

    Bayesian neural controlled differential equations for treatment effect estimation, 2023

    Konstantin Hess, Valentyn Melnychuk, Dennis Frauen, and Stefan Feuerriegel. Bayesian neural controlled differential equations for treatment effect estimation, 2023

    work page 2023

  51. [51]

    Causal modeling of policy interventions from treatment-outcome sequences

    Caglar Hizli, Samuel John, Anne Juuti, Tuure Saarinen, Kirsi Pietilainen, and Pekka Marttinen. Causal modeling of policy interventions from treatment-outcome sequences. In Andreas Krause, Emma Brunskill, Kyunghyun Cho, Barbara Engelhardt, Sivan Sabato, and Jonathan Scarlett, editors,Proceedings of the 40th International Conference on Machine Learning, vol...

    work page 2023

  52. [52]

    Long short-term memory.Neural Computation, 9(8):1735–1780, 1997

    Sepp Hochreiter and Jürgen Schmidhuber. Long short-term memory.Neural Computation, 9(8):1735–1780, 1997

    work page 1997

  53. [53]

    An empirical examination of balancing strategy for counterfactual estimation on time series.arXiv preprint arXiv:2408.08815, 2024

    Qiang Huang, Chuizheng Meng, Defu Cao, Biwei Huang, Yi Chang, and Yan Liu. An empirical examination of balancing strategy for counterfactual estimation on time series.arXiv preprint arXiv:2408.08815, 2024

    work page arXiv 2024

  54. [54]

    Shih-Cheng Huang, Anjul Pareek, Saeed Seyyedi, Imon Banerjee, and Matthew P. Lungren. Fusion of medical imaging and electronic health records using deep learning: a systematic review and implementation guidelines. npj Digital Medicine, 3(1):136, 2020

    work page 2020

  55. [55]

    Causal graph ODE: Continuous treatment effect modeling in multi-agent dynamical systems

    Zijie Huang, Jeehyun Hwang, Junkai Zhang, Jinwoo Baik, Weitong Zhang, Dominik Wodarz, Yizhou Sun, Quan- quan Gu, and Wei Wang. Causal graph ODE: Continuous treatment effect modeling in multi-agent dynamical systems. InProceedings of the ACM Web Conference 2024, pages 4744–4754, 2024. arXiv:2403.00178

    work page arXiv 2024

  56. [56]

    A comprehensive survey of time series foundation models: Understanding, taxonomy, and future directions, 2025

    Ming Jin, Shiyu Wang, Qingsong Long, et al. A comprehensive survey of time series foundation models: Understanding, taxonomy, and future directions, 2025

    work page 2025

  57. [57]

    Neural controlled differential equations for irregular time series

    Patrick Kidger, James Morrill, James Foster, and Terry Lyons. Neural controlled differential equations for irregular time series. InAdvances in Neural Information Processing Systems, 2020

    work page 2020

  58. [58]

    Kingma and Max Welling

    Diederik P. Kingma and Max Welling. Auto-encoding variational bayes. InInternational Conference on Learning Representations, 2014

    work page 2014

  59. [59]

    Kline, H

    A. Kline, H. Wang, Y . Li, et al. Multimodal machine learning in precision health: A scoping review.npj Digital Medicine, 5(1):171, 2022

    work page 2022

  60. [60]

    The artificial intelli- gence clinician learns optimal treatment strategies for sepsis in intensive care.Nature Medicine, 24(11):1716– 1720, 2018

    Matthieu Komorowski, Leo A Celi, Omar Badawi, Anthony C Gordon, and Aldo A Faisal. The artificial intelli- gence clinician learns optimal treatment strategies for sepsis in intensive care.Nature Medicine, 24(11):1716– 1720, 2018

    work page 2018

  61. [61]

    G-net: a deep learning approach to g-computation for counterfactual outcome prediction under dynamic treatment regimes, 2020

    R Li, Z Shahn, J Li, et al. G-net: a deep learning approach to g-computation for counterfactual outcome prediction under dynamic treatment regimes, 2020

    work page 2020

  62. [62]

    BEHRT: Transformer for electronic health records

    Yikuan Li, Shishir Rao, Jose Roberto Ayala Solares, Abdelaali Hassaine, Rema Ramakrishnan, Dexter Canoy, Yajie Zhu, Kazem Rahimi, and Gholamreza Salimi-Khorshidi. BEHRT: Transformer for electronic health records. Scientific Reports, 10:7155, 2020

    work page 2020

  63. [63]

    Eadon, Qianqian Song, Yingjie V

    Zuotian Li, Xiang Liu, Ziyang Tang, Nanxin Jin, Pengyue Zhang, Michael T. Eadon, Qianqian Song, Yingjie V . Chen, and Jing Su. Trajvis: a visual clinical decision support system to translate artificial intelligence trajectory models in the precision management of chronic kidney disease.Journal of the American Medical Informatics Association, 31(11):2474–2...

    work page 2024

  64. [64]

    Alaa, and Mihaela van der Schaar

    Bryan Lim, Ahmed M. Alaa, and Mihaela van der Schaar. Forecasting treatment responses over time using recurrent marginal structural networks.Advances in Neural Information Processing Systems, 31:7494–7504, 2018

    work page 2018

  65. [65]

    Has multimodal learning delivered universal intelligence in healthcare? a comprehensive survey.Information Fusion, 116:102795, 2025

    Qika Lin, Yifan Zhu, Xin Mei, Ling Huang, Jingying Ma, Kai He, Zhen Peng, Erik Cambria, and Mengling Feng. Has multimodal learning delivered universal intelligence in healthcare? a comprehensive survey.Information Fusion, 116:102795, 2025

    work page 2025

  66. [66]

    Estimating individual treatment effects with time-varying confounders

    Ruoqi Liu, Changchang Yin, and Ping Zhang. Estimating individual treatment effects with time-varying confounders. In2020 IEEE International Conference on Data Mining (ICDM), pages 382–391, 2020. 28 arXivTemplate

    work page 2020

  67. [67]

    The neural hawkes process: A neurally self-modulating multivariate point process

    Hongyuan Mei and Jason Eisner. The neural hawkes process: A neurally self-modulating multivariate point process. InAdvances in Neural Information Processing Systems, 2017

    work page 2017

  68. [68]

    Causal transformer for estimating counterfactual outcomes

    Valentyn Melnychuk, Dennis Frauen, and Stefan Feuerriegel. Causal transformer for estimating counterfactual outcomes. In Kamalika Chaudhuri, Stefanie Jegelka, Le Song, Csaba Szepesvari, Gang Niu, and Sivan Sabato, editors,Proceedings of the 39th International Conference on Machine Learning, volume 162 ofProceedings of Machine Learning Research, pages 1529...

    work page 2022

  69. [69]

    Arik, Yan Liu, and Tomas Pfister

    Chuizheng Meng, Yihe Dong, Sercan Ö. Arik, Yan Liu, and Tomas Pfister. Costar: Improved temporal counterfactual estimation with self-supervised learning.arXiv preprint arXiv:2311.00886, abs/2311.00886, 2023

    work page arXiv 2023

  70. [70]

    Kidd, and Joel T

    Riccardo Miotto, Li Li, Brian A. Kidd, and Joel T. Dudley. Deep patient: An unsupervised representation to predict the future of patients from the electronic health records.Scientific Reports, 6:26094, 2016

    work page 2016

  71. [71]

    Krumholz, Jure Leskovec, Eric J

    Michael Moor, Oishi Banerjee, Zahra Shakeri Hossein Abad, Harlan M. Krumholz, Jure Leskovec, Eric J. Topol, and Pranav Rajpurkar. Foundation models for generalist medical artificial intelligence.Nature, 616(7956):259– 265, 2023

    work page 2023

  72. [72]

    EHRWorld: A patient-centric medical world model for long-horizon clinical trajectories

    Linjie Mu, Zhongzhen Huang, Yannian Gu, Shengqian Qin, Shaoting Zhang, and Xiaofan Zhang. Ehrworld: A patient-centric medical world model for long-horizon clinical trajectories.arXiv preprint arXiv:2602.03569, 2026

    work page arXiv 2026

  73. [73]

    Susan A. Murphy. Optimal dynamic treatment regimes.Journal of the Royal Statistical Society: Series B (Statistical Methodology), 65(2):331–355, 2003

    work page 2003

  74. [74]

    Evaluation of trajectory analysis for disease risk assessment: a scoping review.Journal of the American Medical Informatics Association, page ocaf208, 2025

    Freya Pollington, Spiros C Denaxas, Kezhi Li, Johan H Thygesen, Georgios Lyratzopoulos, and Becky White. Evaluation of trajectory analysis for disease risk assessment: a scoping review.Journal of the American Medical Informatics Association, page ocaf208, 2025

    work page 2025

  75. [75]

    D. J. Porche.Epidemiology for the Advanced Practice Nurse: A Population Health Approach. 2022

    work page 2022

  76. [76]

    Hein Putter, Marta Fiocco, and Ronald B. Geskus. Tutorial in biostatistics: competing risks and multi-state models.Statistics in Medicine, 26(11):2389–2430, 2007

    work page 2007

  77. [77]

    Continuous state-space models for optimal sepsis treatment – a deep reinforcement learning approach

    Aniruddh Raghu, Matthieu Komorowski, Leo Anthony Celi, Peter Szolovits, and Marzyeh Ghassemi. Continuous state-space models for optimal sepsis treatment – a deep reinforcement learning approach. In Finale Doshi-Velez, Jim Fackler, Ken Jung, David Kale, Rajesh Ranganath, Byron Wallace, and Jenna Wiens, editors,Proceedings of the 2nd Machine Learning for He...

    work page 2017

  78. [78]

    Continuous state space models for optimal sepsis treatment a deep reinforcement learning approach

    Aniruddh Raghu, Matthieu Komorowski, Leo Anthony Celi, Peter Szolovits, and Marzyeh Ghassemi. Continuous state space models for optimal sepsis treatment a deep reinforcement learning approach. InProceedings of the 2nd Machine Learning for Healthcare Conference, 2017

    work page 2017

  79. [79]

    Rajkomar, E

    A. Rajkomar, E. Oren, K. Chen, et al. Scalable and accurate deep learning with electronic health records.NPJ Digital Medicine, 1(1):18, 2018

    work page 2018

  80. [80]

    Rasmy, Y

    L. Rasmy, Y . Xiang, Z. Xie, et al. Med-bert: pretrained contextualized embeddings on large-scale structured electronic health records for disease prediction.NPJ Digital Medicine, 4(1):86, 2021

    work page 2021

Showing first 80 references.