pith. sign in

arxiv: 2606.24678 · v1 · pith:6YCQNF6Dnew · submitted 2026-06-23 · 📊 stat.ME

Landmarking with Latent Class Mixed Models for Dynamic Prediction of Time-to-event Data with Heterogeneous Biomarker Trajectories

V\'ictor Velasco-Pardo , Nathan Constantine-Cooke , Charlie W. Lees , Catalina A. Vallejos This is my paper

Pith reviewed 2026-06-25 22:25 UTC · model grok-4.3

classification 📊 stat.ME
keywords dynamic predictionlandmarkinglatent class mixed modelstime-to-event databiomarker trajectoriesheterogeneityelectronic health recordssurvival analysis
0
0 comments X

The pith

A landmarking method paired with latent class mixed models captures hidden subgroups in biomarker trajectories to improve dynamic time-to-event predictions.

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

The paper develops a landmarking strategy that incorporates latent class mixed models to handle unobserved subgroups driving differences in how biomarkers change over time. Standard landmarking assumes all patients follow similar longitudinal patterns, an assumption often violated in electronic health record data. By modeling discrete latent classes of trajectories, the approach updates risk estimates as new measurements arrive while remaining fast enough for large datasets. Simulation results show better prediction performance than conventional landmarking when such heterogeneity exists. The method is packaged in landmaRk for flexible use on real-world data.

Core claim

The central claim is that embedding latent class mixed models inside a landmarking framework identifies latent subgroups in longitudinal biomarker data and yields more accurate dynamic predictions of time-to-event outcomes than standard landmarking, without sacrificing computational speed for large-scale applications.

What carries the argument

The integration of latent class mixed models within the landmarking procedure to partition patients into discrete classes based on biomarker trajectory patterns and to condition survival predictions on the assigned class at each landmark time.

If this is right

  • Prediction error decreases relative to homogeneous landmarking when biomarker data contain latent classes.
  • Computation remains feasible on datasets the size of typical electronic health records.
  • The modular implementation allows users to combine the latent-class step with other survival or longitudinal specifications.
  • Risk estimates can be refreshed at each new measurement time while respecting the class structure.

Where Pith is reading between the lines

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

  • The same structure could be tested on other longitudinal covariates such as imaging or wearable sensor streams.
  • Class-specific landmarking might reduce bias in settings where treatment effects differ across unobserved subgroups.
  • Extending the approach to allow time-varying class membership could address cases where patients switch subgroups.

Load-bearing premise

Heterogeneity in biomarker trajectories is produced by a small number of discrete latent subgroups that latent class mixed models can reliably recover.

What would settle it

A simulation study or real dataset in which the proposed method shows no improvement in prediction metrics over standard landmarking when latent heterogeneity is known to be present, or in which the dynamic updating property is lost after class assignment.

Figures

Figures reproduced from arXiv: 2606.24678 by Catalina A. Vallejos, Charlie W. Lees, Nathan Constantine-Cooke, V\'ictor Velasco-Pardo.

Figure 1
Figure 1. Figure 1: Schematic illustration of landmarking with different strategies (LOCF, LME, and LCMM) to summarise the longitudinal [PITH_FULL_IMAGE:figures/full_fig_p003_1.png] view at source ↗
Figure 2
Figure 2. Figure 2: The typical landmaRk pipeline. The landmaRk R package allows analyst to implement landmarking analyses via the following steps: 1) Initialise the landmarking analysis, 2) specify the risk sets (for a pre-specified set of landmark times, 3) fit the longitudinal submodel, 4) fit a survival submodel, and 5) assess the predictive performance. prediction. This was demonstrated e.g., by Tanner and others (2021) … view at source ↗
Figure 3
Figure 3. Figure 3: Ground-truth mean biomarker trajectories for the [PITH_FULL_IMAGE:figures/full_fig_p007_3.png] view at source ↗
Figure 4
Figure 4. Figure 4: Cross-validated tdAUC increment over the static-only model in selected simulation scenarios at landmarks 9 and 12. [PITH_FULL_IMAGE:figures/full_fig_p009_4.png] view at source ↗
Figure 5
Figure 5. Figure 5: Negative increment in cross-validated Brier score over the static-only model in selected simulation scenarios at landmarks [PITH_FULL_IMAGE:figures/full_fig_p010_5.png] view at source ↗
Figure 6
Figure 6. Figure 6: Longitudinal trajectories of square root-transformed CD4 counts in the [PITH_FULL_IMAGE:figures/full_fig_p011_6.png] view at source ↗
Figure 7
Figure 7. Figure 7: Cross-validated increment in tdAUC over the static-only model for the different model specifications in the landmarking [PITH_FULL_IMAGE:figures/full_fig_p013_7.png] view at source ↗
Figure 8
Figure 8. Figure 8: Cross-validated increment in BS over the static-only model for the different model specifications in the landmarking [PITH_FULL_IMAGE:figures/full_fig_p014_8.png] view at source ↗
read the original abstract

The increasing ability to securely access electronic health records (EHR) has created unprecedented opportunities to monitor the health trajectories of large heterogeneous patient populations throughout their lifetime. Repeated measurements of time-varying covariates (such as biomarkers measured via routine blood tests) can inform dynamic risk prediction of time-to-event outcomes, updating risk estimates as new information becomes available. Existing dynamic risk prediction approaches often assume homogeneous longitudinal trajectories across individuals. This assumption is not met when there is heterogeneity driven by latent subgroup structure (e.g. due to unobserved confounders), as is often the case with real-world biomedical data. At present, accounting for such heterogeneity is only available in joint latent class models for longitudinal and time-to-event data, but they are computationally intensive, often prohibitively so for large-scale data, such as those present in EHR settings. To address these challenges, we propose a novel landmarking approach that integrates latent class mixed models (LCMMs) to capture latent heterogeneity in longitudinal trajectories. Our method is implemented in a modular R package, landmaRk, which is available on CRAN and allows users to flexibly specify the components of a landmarking analysis, beyond our proposed approach. Through simulation studies, we demonstrate improvements in prediction performance in the presence of latent heterogeneity compared to traditional landmarking strategies, while remaining computationally efficient for large datasets. We also provide a proof-of-concept illustration using real data.

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

3 major / 2 minor

Summary. The manuscript proposes integrating latent class mixed models (LCMMs) into a landmarking framework to account for latent heterogeneity in longitudinal biomarker trajectories when performing dynamic prediction of time-to-event outcomes. The approach is positioned as computationally efficient for large EHR-scale data relative to joint latent class models, is implemented in the modular landmaRk R package on CRAN, and is supported by simulation studies claiming improved prediction performance over standard landmarking under heterogeneity, plus a real-data illustration.

Significance. If the integration preserves unbiased dynamic updating and the simulation gains hold under realistic identifiability conditions, the method would offer a practical middle ground between homogeneous landmarking and fully joint models, enabling heterogeneity-aware predictions at scale in biomedical settings.

major comments (3)
  1. [§3] §3 (Method): The manuscript must specify whether the LCMM is refitted (or class posteriors re-estimated) using only data observed up to each landmark time t_LM, or whether a single baseline LCMM is used; without this, it is impossible to confirm that the dynamic-updating property of landmarking is retained and that the comparison to traditional landmarking remains valid.
  2. [§4] §4 (Simulations): The reported performance gains rest on the assumption that latent classes remain identifiable and that class probabilities can be conditioned on accumulating longitudinal history without bias; the simulation design should include low-separation and misspecification scenarios to test whether the claimed improvements are robust or artifacts of strong class separation.
  3. [§3.2] §3.2 (Landmarking integration): The paper should provide the explicit form of the landmark-specific Cox model that incorporates the LCMM-derived class probabilities (or random effects), including how any uncertainty in class membership is propagated into the risk predictions; absence of this equation leaves the central claim of unbiased dynamic prediction unverified.
minor comments (2)
  1. The abstract and introduction should cite the specific prior landmarking literature (e.g., the original van Houwelingen or Rizopoulos formulations) against which the new method is compared.
  2. Figure captions in the simulation results should report the exact number of replications and the precise metric definitions (e.g., time-dependent AUC or Brier score) used for the performance comparisons.

Simulated Author's Rebuttal

3 responses · 0 unresolved

We thank the referee for their constructive comments, which help clarify key aspects of our method. We address each major comment below and will revise the manuscript accordingly where revisions are needed.

read point-by-point responses

  1. Referee: §3 (Method): The manuscript must specify whether the LCMM is refitted (or class posteriors re-estimated) using only data observed up to each landmark time t_LM, or whether a single baseline LCMM is used; without this, it is impossible to confirm that the dynamic-updating property of landmarking is retained and that the comparison to traditional landmarking remains valid.

    Authors: We agree that explicit specification is required to verify the dynamic property. In the proposed framework, the LCMM is refitted at each landmark time t_LM using only the longitudinal data observed up to t_LM, with class posteriors re-estimated accordingly before being incorporated into the landmark-specific survival model. This mirrors the dynamic updating in standard landmarking and ensures valid comparisons. We will revise §3 to state this procedure clearly, including pseudocode for the per-landmark fitting step. revision: yes

  2. Referee: §4 (Simulations): The reported performance gains rest on the assumption that latent classes remain identifiable and that class probabilities can be conditioned on accumulating longitudinal history without bias; the simulation design should include low-separation and misspecification scenarios to test whether the claimed improvements are robust or artifacts of strong class separation.

    Authors: The current simulations focus on moderate-to-strong separation to illustrate potential gains when latent heterogeneity is detectable, as is common in proof-of-concept studies. We recognize that robustness under weaker conditions is important. In the revision we will add two new simulation scenarios: (i) low class separation (overlap in trajectories) and (ii) mild misspecification of the number of classes, reporting calibration, discrimination, and computational metrics for all methods. If performance gains diminish, we will qualify the claims accordingly. revision: yes

  3. Referee: §3.2 (Landmarking integration): The paper should provide the explicit form of the landmark-specific Cox model that incorporates the LCMM-derived class probabilities (or random effects), including how any uncertainty in class membership is propagated into the risk predictions; absence of this equation leaves the central claim of unbiased dynamic prediction unverified.

    Authors: We will add the explicit landmark-specific hazard in the revised §3.2: λ(t | t_LM, X, π) = λ0(t) exp(β'X + γ'π), where π denotes the vector of posterior class probabilities obtained from the LCMM fitted up to t_LM. Uncertainty in class membership is propagated by replacing the indicator of class membership with its posterior expectation E[I_c | data up to t_LM] inside the linear predictor; the resulting risk is the posterior-weighted average of class-specific risks. This formulation will be stated mathematically and accompanied by a short derivation showing that the dynamic property is preserved. revision: yes

Circularity Check

0 steps flagged

No significant circularity; method is a new proposal validated externally via simulation

full rationale

The paper proposes integrating LCMMs into landmarking as a computationally efficient alternative to joint latent class models for handling latent heterogeneity in dynamic prediction. The abstract and description frame this as a novel modular approach implemented in the landmaRk R package, with performance claims resting on simulation studies that compare against traditional landmarking (external benchmark) rather than any self-derived quantities. No equations, fitted parameters renamed as predictions, self-citation load-bearing uniqueness theorems, or ansatz smuggling are described. The derivation chain is self-contained against external benchmarks (simulations and real-data proof-of-concept), consistent with the provided reader's assessment of no visible circular reasoning.

Axiom & Free-Parameter Ledger

0 free parameters · 0 axioms · 0 invented entities

Abstract-only review; no specific free parameters, axioms, or invented entities can be identified from the provided text. The approach relies on standard assumptions of mixed models and landmarking but details are absent.

pith-pipeline@v0.9.1-grok · 5798 in / 1146 out tokens · 19972 ms · 2026-06-25T22:25:27.321739+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

87 extracted references · 74 canonical work pages · 1 internal anchor

  1. [1]

    Joint models for longitudinal and time-to-event data:

    Rizopoulos, Dimitris , year =. Joint models for longitudinal and time-to-event data:

  2. [2]

    Position: Stop Chasing the C-index when Evaluating Survival Analysis Models

    Lillelund, Christian Marius and Qi, Shi-ang and Greiner, Russell and Pedersen, Christian Fischer , month = may, year =. Position:. doi:10.48550/arXiv.2506.02075 , abstract =

    work page internal anchor Pith review Pith/arXiv arXiv doi:10.48550/arxiv.2506.02075

  3. [3]

    and Skaloud, J

    Time-. Journal of the American Statistical Association , author =. 2013 , pages =. doi:10.1080/01621459.2013.778776 , language =

    work page doi:10.1080/01621459.2013.778776 2013

  4. [4]

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

    Modelling. Journal of the Royal Statistical Society Series B: Statistical Methodology , author =. 2008 , pages =. doi:10.1111/j.1467-9868.2008.00656.x , abstract =

    work page doi:10.1111/j.1467-9868.2008.00656.x 2008

  5. [5]

    and Gray, Robert J

    A. Journal of the American Statistical Association , author =. 1999 , pages =. doi:10.1080/01621459.1999.10474144 , language =

    work page doi:10.1080/01621459.1999.10474144 1999

  6. [6]

    Biometrics , author =

    Fitting the. Biometrics , author =. 2024 , pages =. doi:10.1093/biomtc/ujae018 , abstract =

    work page doi:10.1093/biomtc/ujae018 2024

  7. [7]

    Clinical Gastroenterology and Hepatology , author =

    Normalization of. Clinical Gastroenterology and Hepatology , author =. 2021 , pages =. doi:10.1016/j.cgh.2020.08.022 , abstract =

    work page doi:10.1016/j.cgh.2020.08.022 2021

  8. [8]

    doi:10.32614/CRAN.package.timeROC , author =

    work page doi:10.32614/cran.package.timeroc

  9. [9]

    Statistics and Its Interface , author =

    A semi-parametric joint latent class model with longitudinal and survival data , volume =. Statistics and Its Interface , author =. 2020 , pages =. doi:10.4310/SII.2020.v13.n3.a10 , language =

    work page doi:10.4310/sii.2020.v13.n3.a10 2020

  10. [10]

    Hepatology , author =

    Primary. Hepatology , author =. 1994 , pages =. doi:10.1002/hep.1840200120 , abstract =

    work page doi:10.1002/hep.1840200120 1994

  11. [11]

    BMC Medical Research Methodology , author =

    Revisiting methods for modeling longitudinal and survival data:. BMC Medical Research Methodology , author =. 2021 , pages =. doi:10.1186/s12874-021-01207-y , abstract =

    work page doi:10.1186/s12874-021-01207-y 2021

  12. [12]

    Soutinho, Gustavo and Meira-Machado, Luis , year =

  13. [13]

    landpred:

    Huynh, Dylan and Parast, Layla , year =. landpred:

  14. [14]

    landmulti:

    Li, Wen and Wang, Qian , year =. landmulti:

  15. [15]

    landmix:

    Garcia, Tanya and Parast, Layla , year =. landmix:

  16. [16]

    landest:

    Parast, Layla , year =. landest:

  17. [17]

    Journal of the Royal Statistical Society Series A: Statistics in Society , author =

    Dynamic. Journal of the Royal Statistical Society Series A: Statistics in Society , author =. 2021 , pages =. doi:10.1111/rssa.12611 , abstract =

    work page doi:10.1111/rssa.12611 2021

  18. [18]

    Dynamic prediction in clinical survival analysis , publisher =

    Van Houwelingen, Hans and Putter, Hein , year =. Dynamic prediction in clinical survival analysis , publisher =

  19. [19]

    Gerds, Thomas Alexander and Ohlendorff, Johan Sebastian and Ozenne, Brice , year =

  20. [20]

    Landmarking:

    Barrott, Isobel and Barrett, Jessica , year =. Landmarking:

  21. [21]

    Regression

    Cox, D R , year =. Regression. Journal of the Royal Statistical Society: Series B (Methodological) , publisher =

  22. [22]

    Rizopoulos, Dimitris and Miranda Afonso, Pedro and Papageorgiou, Grigorios , year =

  23. [23]

    Journal of Open Source Software , author =

    Welcome to the. Journal of Open Source Software , author =. 2019 , pages =. doi:10.21105/joss.01686 , abstract =

    work page doi:10.21105/joss.01686 2019

  24. [24]

    Statistical Methods in Medical Research , author =

    Dynamic prediction by landmarking with data from cohort subsampling designs , issn =. Statistical Methods in Medical Research , author =. 2025 , pages =. doi:10.1177/09622802251403279 , abstract =

    work page doi:10.1177/09622802251403279 2025

  25. [25]

    Kidney International , author =

    Progression of diabetic kidney disease and trajectory of kidney function decline in. Kidney International , author =. 2019 , pages =. doi:10.1016/j.kint.2018.08.026 , language =

    work page doi:10.1016/j.kint.2018.08.026 2019

  26. [26]

    Clinical Gastroenterology and Hepatology , author =

    Large-scale clustering of longitudinal fecal calprotectin and. Clinical Gastroenterology and Hepatology , author =. 2025 , pages =. doi:10.1016/j.cgh.2025.12.022 , language =

    work page doi:10.1016/j.cgh.2025.12.022 2025

  27. [27]

    Frontiers in Neurology , author =

    A comparative study of methods for dynamic survival analysis , volume =. Frontiers in Neurology , author =. 2025 , pages =. doi:10.3389/fneur.2025.1504535 , abstract =

    work page doi:10.3389/fneur.2025.1504535 2025

  28. [28]

    and Sweeting, Michael J

    Barrett, Jessica K. and Sweeting, Michael J. and Wood, Angela M. , year =. Dynamic. Handbook of. doi:10.1016/bs.host.2017.05.004 , abstract =

    work page doi:10.1016/bs.host.2017.05.004 2017

  29. [29]

    Journal of Biomedical Informatics , author =

    A pipeline for harmonising. Journal of Biomedical Informatics , author =. 2025 , pages =. doi:10.1016/j.jbi.2024.104771 , abstract =

    work page doi:10.1016/j.jbi.2024.104771 2025

  30. [30]

    International Journal of Epidemiology , author =

    Reflection on modern methods:. International Journal of Epidemiology , author =. 2021 , pages =. doi:10.1093/ije/dyab047 , abstract =

    work page doi:10.1093/ije/dyab047 2021

  31. [31]

    Clinical Cancer Research , author =

    Conditional. Clinical Cancer Research , author =. 2015 , pages =. doi:10.1158/1078-0432.CCR-14-2154 , abstract =

    work page doi:10.1158/1078-0432.ccr-14-2154 2015

  32. [32]

    Biostatistics , author =

    Fast and flexible inference for joint models of multivariate longitudinal and survival data using integrated nested. Biostatistics , author =. 2024 , pages =. doi:10.1093/biostatistics/kxad019 , abstract =

    work page doi:10.1093/biostatistics/kxad019 2024

  33. [33]

    Biometrical Journal , author =

    Dynamic prediction:. Biometrical Journal , author =. 2020 , pages =. doi:10.1002/bimj.201800248 , abstract =

    work page doi:10.1002/bimj.201800248 2020

  34. [34]

    Circulation: Cardiovascular Quality and Outcomes , author =

    Landmark. Circulation: Cardiovascular Quality and Outcomes , author =. 2011 , pages =. doi:10.1161/CIRCOUTCOMES.110.957951 , abstract =

    work page doi:10.1161/circoutcomes.110.957951 2011

  35. [35]

    The Lancet Digital Health , author =

    Evaluation of performance measures in predictive artificial intelligence models to support medical decisions: overview and guidance , issn =. The Lancet Digital Health , author =. 2025 , pages =. doi:10.1016/j.landig.2025.100916 , language =

    work page doi:10.1016/j.landig.2025.100916 2025

  36. [36]

    Statistics in Medicine , author =

    Assessment and comparison of prognostic classification schemes for survival data , volume =. Statistics in Medicine , author =. 1999 , pages =. doi:10.1002/(SICI)1097-0258(19990915/30)18:17/18<2529::AID-SIM274>3.0.CO;2-5 , abstract =

    work page doi:10.1002/(sici)1097-0258(19990915/30)18:17/18 1999

  37. [37]

    Biostatistics , author =

    The c-index is not proper for the evaluation of \ t\ -year predicted risks , volume =. Biostatistics , author =. 2019 , pages =. doi:10.1093/biostatistics/kxy006 , abstract =

    work page doi:10.1093/biostatistics/kxy006 2019

  38. [38]

    New England Journal of Medicine , author =

    A. New England Journal of Medicine , author =. 1994 , pages =. doi:10.1056/NEJM199403103301001 , abstract =

    work page doi:10.1056/nejm199403103301001 1994

  39. [39]

    The American Statistician , author =

    Separate and. The American Statistician , author =. 2004 , pages =. doi:10.1198/0003130042854 , language =

    work page doi:10.1198/0003130042854 2004

  40. [40]

    Burns & Trauma , author =

    Serial platelet count as a dynamic prediction marker of hospital mortality among septic patients , volume =. Burns & Trauma , author =. 2024 , pages =. doi:10.1093/burnst/tkae016 , abstract =

    work page doi:10.1093/burnst/tkae016 2024

  41. [41]

    BMC Medical Research Methodology , author =

    External validation of a. BMC Medical Research Methodology , author =. 2013 , pages =. doi:10.1186/1471-2288-13-33 , abstract =

    work page doi:10.1186/1471-2288-13-33 2013

  42. [42]

    Statistics in Medicine , author =

    Landmarking 2.0:. Statistics in Medicine , author =. 2022 , pages =. doi:10.1002/sim.9336 , abstract =

    work page doi:10.1002/sim.9336 2022

  43. [43]

    Health Care Management Science , author =

    Using the landmark method for creating prediction models in large datasets derived from electronic health records , volume =. Health Care Management Science , author =. 2015 , pages =. doi:10.1007/s10729-014-9281-3 , abstract =

    work page doi:10.1007/s10729-014-9281-3 2015

  44. [44]

    Health Affairs , author =

    Big. Health Affairs , author =. 2014 , pages =. doi:10.1377/hlthaff.2014.0041 , abstract =

    work page doi:10.1377/hlthaff.2014.0041 2014

  45. [45]

    PLOS One , author =

    Longitudinal modeling of red blood cell distribution width dynamics and mortality risk in critically. PLOS One , author =. 2025 , pages =. doi:10.1371/journal.pone.0333605 , abstract =

    work page doi:10.1371/journal.pone.0333605 2025

  46. [46]

    International Journal of Bio-Medical Computing , author =

    Calculating confidence intervals for summary measures of individual curves via nonlinear regression models , volume =. International Journal of Bio-Medical Computing , author =. 1996 , pages =. doi:10.1016/0020-7101(95)01152-8 , abstract =

    work page doi:10.1016/0020-7101(95)01152-8 1996

  47. [47]

    Statistics in Medicine , author =

    A refinement to the analysis of serial data using summary measures , volume =. Statistics in Medicine , author =. 1993 , pages =. doi:10.1002/sim.4780120105 , abstract =

    work page doi:10.1002/sim.4780120105 1993

  48. [48]

    Analysis of serial measurements in medical research , volume =

    Matthews, J N S and Altman, Douglas G and Campbell, M J and Royston, Patrick , year =. Analysis of serial measurements in medical research , volume =

  49. [49]

    and Prentice, Ross L

    Kalbfleisch, John D. and Prentice, Ross L. , month = aug, year =. The. doi:10.1002/9781118032985 , language =

    work page doi:10.1002/9781118032985

  50. [50]

    Statistics in Medicine , author =

    Discovering subpopulation structure with latent class mixed models , volume =. Statistics in Medicine , author =. 2002 , pages =. doi:10.1002/sim.1027 , abstract =

    work page doi:10.1002/sim.1027 2002

  51. [51]

    Biometrics , author =

    A. Biometrics , author =. 1997 , pages =. doi:10.2307/2533118 , abstract =

    work page doi:10.2307/2533118 1997

  52. [52]

    Integrating latent classes in the

    Andrinopoulou, Eleni-Rosalina and Nasserinejad, Kazem and Szczesniak, Rhonda and Rizopoulos, Dimitris , month = nov, year =. Integrating latent classes in the. Statistical Methods in Medical Research , publisher =. doi:10.1177/0962280220924680 , abstract =

    work page doi:10.1177/0962280220924680

  53. [53]

    BMC Medical Research Methodology , author =

    Joint latent class model:. BMC Medical Research Methodology , author =. 2021 , pages =. doi:10.1186/s12874-021-01377-9 , abstract =

    work page doi:10.1186/s12874-021-01377-9 2021

  54. [54]

    and McLean, Colin and Hall, Peter S

    Sierra, Begoña B. and McLean, Colin and Hall, Peter S. and Vallejos, Catalina A. , year =. The. doi:10.48550/ARXIV.2508.14821 , abstract =

    work page doi:10.48550/arxiv.2508.14821

  55. [55]

    British Journal of Mathematical and Statistical Psychology , author =

    Analysis of multivariate mixed longitudinal data:. British Journal of Mathematical and Statistical Psychology , author =. 2013 , pages =. doi:10.1111/bmsp.12000 , abstract =

    work page doi:10.1111/bmsp.12000 2013

  56. [56]

    Biometrics , author =

    A. Biometrics , author =. 2006 , pages =. doi:10.1111/j.1541-0420.2006.00573.x , abstract =

    work page doi:10.1111/j.1541-0420.2006.00573.x 2006

  57. [57]

    Gastroenterology , author =

    Disease. Gastroenterology , author =. 2022 , pages =. doi:10.1053/j.gastro.2022.01.024 , language =

    work page doi:10.1053/j.gastro.2022.01.024 2022

  58. [58]

    , month = oct, year =

    Chen, Sida and Alvares, Danilo and Palma, Marco and Barrett, Jessica K. , month = oct, year =. Bayesian shared parameter joint models for heterogeneous populations , volume =. Statistics and Computing , publisher =. doi:10.1007/s11222-025-10647-1 , abstract =

    work page doi:10.1007/s11222-025-10647-1

  59. [59]

    Clinical Gastroenterology and Hepatology , author =

    Longitudinal. Clinical Gastroenterology and Hepatology , author =. 2023 , pages =. doi:10.1016/j.cgh.2023.03.026 , language =

    work page doi:10.1016/j.cgh.2023.03.026 2023

  60. [60]

    Frontiers in Big Data , author =

    Disease. Frontiers in Big Data , author =. 2021 , pages =. doi:10.3389/fdata.2021.676168 , abstract =

    work page doi:10.3389/fdata.2021.676168 2021

  61. [61]

    Epidemiology , author =

    Prediction. Epidemiology , author =. 2024 , pages =. doi:10.1097/EDE.0000000000001713 , abstract =

    work page doi:10.1097/ede.0000000000001713 2024

  62. [62]

    JAMA , author =

    Individualized. JAMA , author =. doi:10.1001/jama.2025.14194 , abstract =

    work page doi:10.1001/jama.2025.14194 2025

  63. [63]

    Journal of Biomedical Informatics , author =

    Soft phenotyping for sepsis via. Journal of Biomedical Informatics , author =. 2024 , pages =. doi:10.1016/j.jbi.2024.104615 , abstract =

    work page doi:10.1016/j.jbi.2024.104615 2024

  64. [64]

    2023 , pages =

    Scientific Data , author =. 2023 , pages =. doi:10.1038/s41597-022-01899-x , abstract =

    work page doi:10.1038/s41597-022-01899-x 2023

  65. [65]

    2024 , pages =

    Journal of the American Medical Informatics Association , author =. 2024 , pages =. doi:10.1093/jamia/ocae271 , abstract =

    work page doi:10.1093/jamia/ocae271 2024

  66. [66]

    American Journal of Epidemiology , author =

    Landmark. American Journal of Epidemiology , author =. 2018 , pages =. doi:10.1093/aje/kwy018 , language =

    work page doi:10.1093/aje/kwy018 2018

  67. [67]

    , language =

    Analysis of survival by tumor response. , language =. Journal of clinical oncology , author =

  68. [68]

    Biometrics , author =

    Modelling. Biometrics , author =. 1994 , pages =. doi:10.2307/2533439 , abstract =

    work page doi:10.2307/2533439 1994

  69. [69]

    Journal of the American Statistical Association , author =

    Modeling. Journal of the American Statistical Association , author =. 1993 , pages =. doi:10.2307/2290756 , language =

    work page doi:10.2307/2290756 1993

  70. [70]

    Statistics in Medicine , author =

    Dynamic evaluation of short‐term prognosis after myocardial infarction , volume =. Statistics in Medicine , author =. 1985 , pages =. doi:10.1002/sim.4780040106 , abstract =

    work page doi:10.1002/sim.4780040106 1985

  71. [71]

    The American Journal of Cardiology , author =

    Dynamic evaluation of prognosis from time-dependent variables in acute myocardial infarction , volume =. The American Journal of Cardiology , author =. 1983 , pages =. doi:10.1016/0002-9149(83)90190-X , language =

    work page doi:10.1016/0002-9149(83)90190-x 1983

  72. [72]

    2018 , pages =

    BMC Medical Research Methodology , author =. 2018 , pages =. doi:10.1186/s12874-018-0482-1 , abstract =

    work page doi:10.1186/s12874-018-0482-1 2018

  73. [73]

    BMC Medical Research Methodology , author =

    Random survival forests for dynamic predictions of a time-to-event outcome using a longitudinal biomarker , volume =. BMC Medical Research Methodology , author =. 2021 , pages =. doi:10.1186/s12874-021-01375-x , abstract =

    work page doi:10.1186/s12874-021-01375-x 2021

  74. [74]

    and Williamson, Paula and Kolamunnage-Dona, Ruwanthi and Henderson, Robin and Hickey, Graeme L

    Philipson, Pete and Sousa, Ines and Diggle, Peter J. and Williamson, Paula and Kolamunnage-Dona, Ruwanthi and Henderson, Robin and Hickey, Graeme L. , month = mar, year =. doi:10.32614/CRAN.package.joineR , abstract =

    work page doi:10.32614/cran.package.joiner

  75. [75]

    Journal of Statistical Software , author =

    The. Journal of Statistical Software , author =. doi:10.18637/jss.v072.i07 , abstract =

    work page doi:10.18637/jss.v072.i07

  76. [76]

    doi:10.18637/jss.v035.i09 , abstract =

    Journal of Statistical Software , author =. doi:10.18637/jss.v035.i09 , abstract =

    work page doi:10.18637/jss.v035.i09

  77. [77]

    Journal of Statistical Software , author =

    Estimation of. Journal of Statistical Software , author =. doi:10.18637/jss.v078.i02 , abstract =

    work page doi:10.18637/jss.v078.i02

  78. [78]

    Journal of the American Statistical Association , author =

    Latent. Journal of the American Statistical Association , author =. 2002 , pages =. doi:10.1198/016214502753479220 , language =

    work page doi:10.1198/016214502753479220 2002

  79. [79]

    Statistics in Medicine , author =

    A latent class mixed model for analysing biomarker trajectories with irregularly scheduled observations , volume =. Statistics in Medicine , author =. 2000 , pages =. doi:10.1002/(SICI)1097-0258(20000530)19:10<1303::AID-SIM424>3.0.CO;2-E , abstract =

    work page doi:10.1002/(sici)1097-0258(20000530)19:10 2000

  80. [80]

    Computational Statistics & Data Analysis , author =

    Joint modelling of multivariate longitudinal outcomes and a time-to-event:. Computational Statistics & Data Analysis , author =. 2009 , pages =. doi:10.1016/j.csda.2008.10.017 , abstract =

    work page doi:10.1016/j.csda.2008.10.017 2009

Showing first 80 references.