pith. sign in

arxiv: 2605.03633 · v1 · submitted 2026-05-05 · 📊 stat.ME

Variable Domain Multivariate Functional Principal Component Analysis

Pavel Hern\'andez Amaro , Mar\'ia Durb\'an , M. Carmen Aguilera-Morillo , Jos\'e Mar\'ia Quintana , Irantzu Barrio , Sonja Greven This is my paper

Pith reviewed 2026-05-07 13:48 UTC · model grok-4.3

classification 📊 stat.ME
keywords variable domainmultivariate functional PCAfunctional data analysisdimension reductioneigenfunction estimationlongitudinal trajectories
0
0 comments X

The pith

Multivariate functional principal component analysis can be extended to data recorded over subject-specific intervals.

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

The paper develops an approach to multivariate functional principal component analysis when each subject's observations span a unique interval length instead of a shared fixed domain. This matters in applications such as medical monitoring, where patients may be tracked for different durations and standard methods that force a common grid can distort the joint patterns across variables. The procedure computes univariate variable-domain functional principal components for each variable on its own, stacks those scores, and then smooths the empirical covariance matrix of the stacked scores with respect to domain length. Simulations indicate that the resulting multivariate eigenfunctions and scores recover the underlying structure more accurately than methods that ignore variable lengths or rely on binning. The approach is illustrated on body temperature and oxygen saturation trajectories from COVID-19 patients whose hospital stays produced unequal monitoring periods.

Core claim

Existing MFPCA methods assume all functional observations are recorded over a common, fixed domain. To accommodate variable domains, we perform univariate variable domain FPCA for each functional variable separately, stack the resulting univariate scores, and smooth the empirical covariance matrix of these stacked scores over the domain length. This yields estimates of multivariate eigenfunctions and scores that properly account for varying observation periods.

What carries the argument

Stacking of univariate variable-domain FPCA scores whose empirical covariance is then smoothed over domain length to recover joint eigenfunctions.

If this is right

  • Multivariate eigenfunctions and scores estimated this way correctly incorporate information from subjects observed over shorter or longer periods.
  • The estimates avoid the information loss that occurs when data are forced onto a common grid through binning or truncation.
  • Joint variation across multiple trajectories can be extracted from longitudinal records that have unequal lengths, as shown in the temperature and SpO2 example.

Where Pith is reading between the lines

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

  • The stacking step after separate univariate processing could be inserted into other multivariate functional techniques such as regression or clustering that also need to respect variable domains.
  • The method points toward hybrids with registration or warping approaches that might further align patterns when both domain lengths and internal timing vary.

Load-bearing premise

Separately computed univariate scores retain enough cross-variable dependence that stacking them and smoothing their covariance produces accurate joint multivariate components.

What would settle it

Simulate multivariate functional data with known true eigenstructure and deliberately varying domain lengths per subject, then measure whether the method recovers the true eigenfunctions and scores with error rates comparable to an oracle that uses the full joint information.

read the original abstract

Multivariate functional principal component analysis (MFPCA) is a powerful dimension reduction technique for analyzing multiple functional variables simultaneously. However, existing MFPCA methods assume that all functional observations are recorded over a common, fixed domain. This assumption is often violated in practical applications where the observation period varies across subjects, leading to what is known as variable domain functional data. We propose a novel approach for MFPCA that explicitly accommodates variable domains by extending existing multivariate functional principal component analysis to the variable domain setting. Our methodology involves performing univariate variable domain FPCA for each functional variable separately, stacking the resulting univariate scores, and then smoothing the empirical covariance matrix of these stacked scores over the domain length. This allows us to estimate multivariate eigenfunctions and scores that properly account for varying observation periods. We demonstrate through extensive simulation studies that our proposed method outperforms approaches that ignore the variable domain structure and rely on binning strategies. The practical utility of our method is illustrated through an application analyzing body temperature and capillary oxygen saturation (SpO$_2$) trajectories in COVID-19 hospital admitted patients, where patients experienced varying lengths of stay and monitoring periods.

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 / 3 minor

Summary. The manuscript proposes a novel MFPCA procedure for functional data observed over subject-specific domains. It proceeds by first computing univariate variable-domain FPCA scores separately for each functional variable, stacking these scores, smoothing the resulting empirical covariance matrix with respect to observed domain length, and finally extracting multivariate eigenfunctions and scores from the smoothed matrix. The authors claim that this construction yields valid joint components that properly account for varying observation intervals, demonstrate superiority over binning-based competitors in simulation studies, and illustrate the method on COVID-19 patient trajectories of body temperature and SpO2.

Significance. If the procedure is shown to be consistent, it supplies a practical, computationally straightforward extension of existing two-step MFPCA pipelines to the variable-domain setting that arises frequently in longitudinal biomedical data. The stacking step preserves cross-variable information through the off-block entries of the score covariance, and the subsequent smoothing acts as a natural regularizer when the number of retained scores varies with domain length. The reported simulation gains and the real-data application to irregular hospital monitoring periods indicate immediate utility, although the lack of accompanying theoretical results (consistency, rates, or bias bounds) tempers the overall contribution.

major comments (2)
  1. [Methodology] Methodology section (after the description of stacking univariate scores): the smoothed covariance estimator is introduced without an explicit formula, without specification of the kernel or bandwidth selector, and without analysis of how smoothing over domain length affects the eigen-decomposition of the joint operator. Because this step is load-bearing for the multivariate eigenfunctions, its properties should be stated precisely.
  2. [Simulation studies] Simulation studies section, performance tables: the reported superiority is quantified only by aggregate error measures; the manuscript does not state the number of Monte Carlo replications, does not report variability across replications, and does not vary the distribution of domain lengths in a controlled way. These omissions make it difficult to judge whether the gains are robust or merely artifacts of the chosen simulation design.
minor comments (3)
  1. [Abstract and Introduction] The abstract and introduction refer to 'extensive simulation studies' but the main text provides only high-level descriptions of the data-generating process; a concise algorithmic summary or pseudocode for the full procedure would improve reproducibility.
  2. [Methodology] Notation for the stacked score vector and the domain-length smoothing operator is introduced informally; explicit definitions (e.g., an equation for the smoothed matrix) would eliminate ambiguity when readers attempt to implement the method.
  3. [Application] The real-data application section would benefit from a brief sensitivity check showing how results change with the number of retained univariate scores or with different smoothing parameters.

Simulated Author's Rebuttal

2 responses · 0 unresolved

We thank the referee for the constructive comments on our manuscript. We address each major comment below and have revised the manuscript accordingly to improve clarity and completeness.

read point-by-point responses

  1. Referee: [Methodology] Methodology section (after the description of stacking univariate scores): the smoothed covariance estimator is introduced without an explicit formula, without specification of the kernel or bandwidth selector, and without analysis of how smoothing over domain length affects the eigen-decomposition of the joint operator. Because this step is load-bearing for the multivariate eigenfunctions, its properties should be stated precisely.

    Authors: We agree that the smoothed covariance step requires a more precise description. In the revised manuscript we will insert the explicit formula for the smoothed estimator (a kernel-weighted average of the empirical score covariance matrices, indexed by observed domain length), specify the kernel (Gaussian) and bandwidth selector (leave-one-out cross-validation on domain length), and add a short paragraph explaining that the smoothing regularizes the joint covariance operator when the number of retained univariate scores varies with domain length, thereby preserving the off-block cross-variable information while stabilizing the subsequent eigen-decomposition. revision: yes

  2. Referee: [Simulation studies] Simulation studies section, performance tables: the reported superiority is quantified only by aggregate error measures; the manuscript does not state the number of Monte Carlo replications, does not report variability across replications, and does not vary the distribution of domain lengths in a controlled way. These omissions make it difficult to judge whether the gains are robust or merely artifacts of the chosen simulation design.

    Authors: We acknowledge these omissions. The revised simulation section will explicitly state the number of Monte Carlo replications performed, report standard errors or inter-quartile ranges of the error measures across replications in the tables, and include an additional controlled experiment that varies the distribution of domain lengths (e.g., uniform versus right-skewed) while keeping other factors fixed, to demonstrate that the reported gains are not artifacts of the original design. revision: yes

Circularity Check

0 steps flagged

No significant circularity; method is a direct, non-reductive extension of standard MFPCA

full rationale

The paper's core procedure—univariate variable-domain FPCA per margin, stacking of scores, kernel smoothing of the finite-dimensional score covariance over observed domain length, followed by eigen-decomposition—directly implements the classical two-step MFPCA pipeline on preprocessed scores. No equation defines a multivariate eigenfunction or score in terms of itself, no fitted parameter is relabeled as a prediction of the same quantity, and no load-bearing uniqueness result is imported from the authors' prior work. The cross-covariance structure is preserved in the off-block entries of the stacked covariance, so the subsequent decomposition incorporates joint information without algebraic reduction to the inputs. This is a standard methodological extension whose validity rests on external statistical properties of FPCA and smoothing, not on self-referential construction.

Axiom & Free-Parameter Ledger

1 free parameters · 1 axioms · 0 invented entities

The method rests on the validity of univariate variable-domain FPCA applied marginally and on the assumption that smoothing the stacked-score covariance recovers the joint structure; no new entities are postulated and the smoothing parameter is an implicit free choice.

free parameters (1)
  • smoothing parameter for the stacked-score covariance matrix
    Smoothing the empirical covariance over domain length requires a bandwidth or penalty parameter whose selection is not specified in the abstract.
axioms (1)
  • domain assumption Univariate variable-domain FPCA produces valid scores for each functional variable independently
    The pipeline begins by applying univariate variable-domain FPCA separately to each margin before stacking.

pith-pipeline@v0.9.0 · 5519 in / 1423 out tokens · 69803 ms · 2026-05-07T13:48:52.633480+00:00 · methodology

Review history (2 revisions) →

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

84 extracted references · 84 canonical work pages

  1. [1]

    Chen and G

    Zhou, Rensheng R. and Serban, Nicoleta and Gebraeel, Nagi and M. 2014 , journal =. doi:10.1080/00401706.2013.805661 , issn =

    work page doi:10.1080/00401706.2013.805661 2014

  2. [2]

    2017 , journal =

    Greven, Sonja and Scheipl, Fabian , number =. 2017 , journal =. doi:10.1177/1471082X16681317 , issn =

    work page doi:10.1177/1471082x16681317 2017

  3. [3]

    2013 , journal =

    Ivanescu, Andrada , number =. 2013 , journal =. doi:10.5539/ijsp.v2n2p102 , issn =

    work page doi:10.5539/ijsp.v2n2p102 2013

  4. [4]

    2011 , journal =

    Chen, Huaihou and Wang, Yuanjia , number =. 2011 , journal =. doi:10.1111/j.1541-0420.2010.01524.x , issn =

    work page doi:10.1111/j.1541-0420.2010.01524.x 2011

  5. [5]

    , publisher =

    De Boor, C. , publisher =. 2001 , booktitle =

    work page 2001

  6. [6]

    2014 , booktitle =

    Shang, Han Lin , number =. 2014 , booktitle =. doi:10.1007/s10182-013-0213-1 , issn =

    work page doi:10.1007/s10182-013-0213-1 2014

  7. [7]

    An introduction to ROC analysis

    Fawcett, Tom , number =. 2006 , journal =. doi:10.1016/j.patrec.2005.10.010 , issn =

    work page doi:10.1016/j.patrec.2005.10.010 2006

  8. [8]

    2018 , journal =

    Bauer, Alexander and Scheipl, Fabian and K. 2018 , journal =. doi:10.1177/1471082X17748034 , issn =

    work page doi:10.1177/1471082x17748034 2018

  9. [9]

    2014 , journal =

    Zhang, Chuanrong and Li, Weidong and Civco, Daniel , number =. 2014 , journal =. doi:10.1080/01431161.2014.975377 , issn =

    work page doi:10.1080/01431161.2014.975377 2014

  10. [10]

    1993 , journal =

    Breslow, N E and Clayton, D G , number =. 1993 , journal =

    work page 1993

  11. [11]

    and Hall, P

    Delaigle, A. and Hall, P. , number =. 2016 , journal =. doi:10.1093/biomet/asw040 , issn =

    work page doi:10.1093/biomet/asw040 2016

  12. [12]

    Zhenhua, L. I.N. and Wang, Jane Ling and Zhong, Qixian , number =. 2021 , journal =. doi:10.1093/biomet/asaa088 , issn =

    work page doi:10.1093/biomet/asaa088 2021

  13. [13]

    2010 , journal =

    Guillas, Serge and Lai, Ming Jun , number =. 2010 , journal =. doi:10.1080/10485250903323180 , issn =

    work page doi:10.1080/10485250903323180 2010

  14. [14]

    2017 , journal =

    Brockhaus, Sarah and Melcher, Michael and Leisch, Friedrich and Greven, Sonja , number =. 2017 , journal =. doi:10.1007/s11222-016-9662-1 , issn =

    work page doi:10.1007/s11222-016-9662-1 2017

  15. [15]

    Applied Computing and Infor- matics 17(1), 168–192

    Tharwat, Alaa , number =. 2018 , journal =. doi:10.1016/j.aci.2018.08.003 , issn =

    work page doi:10.1016/j.aci.2018.08.003 2018

  16. [16]

    2013 , journal =

    Delaigle, Aurore and Hall, Peter , number =. 2013 , journal =. doi:10.1080/01621459.2013.824893 , issn =

    work page doi:10.1080/01621459.2013.824893 2013

  17. [17]

    2023 , journal =

    Portuondo-Jim. 2023 , journal =

    work page 2023

  18. [18]

    Aguilera, A. M. and Aguilera-Morillo, M. C. , number =. 2013 , journal =. doi:10.1016/j.mcm.2013.04.007 , issn =

    work page doi:10.1016/j.mcm.2013.04.007 2013

  19. [19]

    2015 , journal =

    Kraus, David , number =. 2015 , journal =. doi:10.1111/rssb.12087 , issn =

    work page doi:10.1111/rssb.12087 2015

  20. [20]

    2013 , journal =

    Lee, Dae Jin and Durb. 2013 , journal =. doi:10.1016/j.csda.2012.11.013 , issn =

    work page doi:10.1016/j.csda.2012.11.013 2013

  21. [21]

    2005 , journal =

    Cardot, Hervé and Sarda, Pacal , number =. 2005 , journal =. doi:10.1016/j.jmva.2003.08.008 , issn =

    work page doi:10.1016/j.jmva.2003.08.008 2005

  22. [22]

    2017 , journal =

    Gromenko, Oleksandr and Kokoszka, Piotr and Sojka, Jan , number =. 2017 , journal =. doi:10.1214/17-AOAS1022 , issn =

    work page doi:10.1214/17-aoas1022 2017

  23. [23]

    and Currie, Iain D

    Eilers, Paul H.C. and Currie, Iain D. and Durb. 2006 , journal =. doi:10.1016/j.csda.2004.07.008 , issn =

    work page doi:10.1016/j.csda.2004.07.008 2006

  24. [24]

    Militino, A F · and Ugarte, Montesino M and Militino, A F and Montesino, M , keywords =

    work page

  25. [25]

    1996 , booktitle =

    Eilers, Paul H C and Marx, Brian D , number =. 1996 , booktitle =

    work page 1996

  26. [26]

    , number =

    Ivanescu, Andrada E. , number =. 2018 , journal =. doi:10.1080/03610918.2017.1353619 , issn =

    work page doi:10.1080/03610918.2017.1353619 2018

  27. [27]

    2003 , journal =

    Cardot, Hervé and Faivre, Robert and Goulard, Michel , number =. 2003 , journal =. doi:10.1080/0266476032000107187 , issn =

    work page doi:10.1080/0266476032000107187 2003

  28. [28]

    Yao, H.-G

    Yao, Fang and M. 2005 , journal =. doi:10.1198/016214504000001745 , issn =

    work page doi:10.1198/016214504000001745 2005

  29. [29]

    2009 , booktitle =

    Ramsay, James and Hooker, Giles and Graves, Spencer , publisher =. 2009 , booktitle =

    work page 2009

  30. [30]

    doi:10.1007/978-3-031-08329-7

    2022 , author =. doi:10.1007/978-3-031-08329-7

    work page doi:10.1007/978-3-031-08329-7 2022

  31. [31]

    and Ogden, R

    Reiss, Philip T. and Ogden, R. Todd , number =. 2007 , journal =. doi:10.1198/016214507000000527 , issn =

    work page doi:10.1198/016214507000000527 2007

  32. [32]

    and Shen, Haipeng and Buja, Andreas , pages =

    Huang, Jianhua Z. and Shen, Haipeng and Buja, Andreas , pages =. 2008 , journal =. doi:10.1214/08-EJS218 , issn =

    work page doi:10.1214/08-ejs218 2008

  33. [33]

    2005 , booktitle =

    M. 2005 , booktitle =. doi:10.1214/009053604000001156 , issn =

    work page doi:10.1214/009053604000001156 2005

  34. [34]

    1971 , journal =

    Rao, C Radhakrishna and Mitra, S K , number =. 1971 , journal =

    work page 1971

  35. [35]

    2017 , journal =

    Wang, Xiao and Zhu, Hongtu , number =. 2017 , journal =. doi:10.1080/01621459.2016.1194846 , issn =

    work page doi:10.1080/01621459.2016.1194846 2017

  36. [36]

    2013 , journal =

    Pruijm, Menno and Ponte, Belen and Ackermann, Daniel and Vuistiner, Philippe and Paccaud, Fred and Guessous, Idris and Ehret, Georg and Eisenberger, Ute and Mohaupt, Markus and Burnier, Michel and Martin, Pierre Yves and Bochud, Murielle , number =. 2013 , journal =. doi:10.1007/s00330-013-2900-4 , issn =

    work page doi:10.1007/s00330-013-2900-4 2013

  37. [37]

    2025 , journal =

    Rodriguez-Idiazabal, Lander and Fern. 2025 , journal =. doi:10.1186/s13690-025-01681-6 , issn =

    work page doi:10.1186/s13690-025-01681-6 2025

  38. [38]

    1999 , booktitle =

    Lin, Xihong and Zhang, Daowen , number =. 1999 , booktitle =

    work page 1999

  39. [39]

    2019 , journal =

    Kraus, David , month =. 2019 , journal =. doi:10.1016/j.jmva.2019.05.002 , issn =

    work page doi:10.1016/j.jmva.2019.05.002 2019

  40. [40]

    and Clocher, Ilona C

    Kulbaba, Mason W. and Clocher, Ilona C. and Harder, Lawrence D. , number =. 2017 , journal =. doi:10.1111/jse.12252 , issn =

    work page doi:10.1111/jse.12252 2017

  41. [41]

    and Aizpiri, Susana and Basualdo, Luis V

    Esteban, Cristóbal and Arostegui, Inmaculada and Aburto, Myriam and Moraza, Javier and Quintana, José M. and Aizpiri, Susana and Basualdo, Luis V. and Capelastegui, Alberto , number =. 2014 , journal =. doi:10.1111/resp.12239 , issn =

    work page doi:10.1111/resp.12239 2014

  42. [42]

    and Nuzzo, Regina L

    Levitin, Daniel J. and Nuzzo, Regina L. and Vines, Bradleyw and Ramsay, J. O. , number =. 2007 , booktitle =. doi:10.1037/cp2007014 , issn =

    work page doi:10.1037/cp2007014 2007

  43. [43]

    2019 , journal =

    Wang, Wei and Fang, Zhuo , number =. 2019 , journal =. doi:10.1080/02664763.2018.1502262 , issn =

    work page doi:10.1080/02664763.2018.1502262 2019

  44. [44]

    2009 , journal =

    Christensen, Erik , number =. 2009 , journal =

    work page 2009

  45. [45]

    doi:10.2436/20.8080.02.124 , issn =

    2019 , author =. doi:10.2436/20.8080.02.124 , issn =

    work page doi:10.2436/20.8080.02.124 2019

  46. [46]

    2014 , journal =

    Jacques, Julien and Preda, Cristian , month =. 2014 , journal =. doi:10.1016/J.CSDA.2012.12.004 , issn =

    work page doi:10.1016/j.csda.2012.12.004 2014

  47. [47]

    2022 , journal =

    Gaynanova, Irina and Punjabi, Naresh and Crainiceanu, Ciprian , number =. 2022 , journal =. doi:10.1093/biostatistics/kxaa023 , issn =

    work page doi:10.1093/biostatistics/kxaa023 2022

  48. [48]

    and Eilers, Paul H.C

    Marx, Brian D. and Eilers, Paul H.C. , number =. 2005 , journal =. doi:10.1198/004017004000000626 , issn =

    work page doi:10.1198/004017004000000626 2005

  49. [49]

    2017 , journal =

    Happ, Clara and Greven, Sonja , number =. 2018 , journal =. doi:10.1080/01621459.2016.1273115 , issn =

    work page doi:10.1080/01621459.2016.1273115 2018

  50. [50]

    2019 , journal =

    Rodr. 2019 , journal =. doi:10.1007/s11222-018-9818-2 , issn =

    work page doi:10.1007/s11222-018-9818-2 2019

  51. [51]

    2020 , journal =

    Kneip, Alois and Liebl, Dominik , number =. 2020 , journal =. doi:10.1214/19-AOS1864 , issn =

    work page doi:10.1214/19-aos1864 2020

  52. [52]

    and Aburto, Myriam and Capelastegui, Alberto , number =

    Esteban, Cristóbal and Moraza, Javier and Iriberri, Milagros and Aguirre, Urko and Goiria, Begoña and Quintana, José M. and Aburto, Myriam and Capelastegui, Alberto , number =. 2016 , journal =. doi:10.2147/COPD.S115350 , issn =

    work page doi:10.2147/copd.s115350 2016

  53. [53]

    , number =

    Wood, Simon N. , number =. 2017 , journal =. doi:10.1007/s11222-016-9666-x , issn =

    work page doi:10.1007/s11222-016-9666-x 2017

  54. [54]

    2019 , journal =

    Liebl, Dominik and Rameseder, Stefan , month =. 2019 , journal =. doi:10.1016/j.csda.2018.08.011 , issn =

    work page doi:10.1016/j.csda.2018.08.011 2019

  55. [55]

    and Brutti, Pierpaolo , number =

    Stefanucci, Marco and Sangalli, Laura M. and Brutti, Pierpaolo , number =. 2018 , journal =. doi:10.1111/stan.12137 , issn =

    work page doi:10.1111/stan.12137 2018

  56. [56]

    and Caffo, Brian and Reich, Daniel , number =

    Goldsmith, Jeff and Bobb, Jennifer and Crainiceanu, Ciprian M. and Caffo, Brian and Reich, Daniel , number =. 2011 , journal =. doi:10.1198/jcgs.2010.10007 , issn =

    work page doi:10.1198/jcgs.2010.10007 2011

  57. [57]

    2017 , journal =

    Cai, Zhanzhang and J. 2017 , journal =. doi:10.3390/rs9121271 , issn =

    work page doi:10.3390/rs9121271 2017

  58. [58]

    and Schaepman-Strub, Gabriela and Furrer, Reinhard , number =

    Gerber, Florian and De Jong, Rogier and Schaepman, Michael E. and Schaepman-Strub, Gabriela and Furrer, Reinhard , number =. 2018 , journal =. doi:10.1109/TGRS.2017.2785240 , issn =

    work page doi:10.1109/tgrs.2017.2785240 2018

  59. [59]

    and Ritov, Y

    Goldberg, Y. and Ritov, Y. and Mandelbaum, A. , month =. 2014 , journal =. doi:10.1016/j.jspi.2013.11.006 , issn =

    work page doi:10.1016/j.jspi.2013.11.006 2014

  60. [60]

    Carmen and Durb

    Aguilera-Morillo, M. Carmen and Durb. 2017 , journal =. doi:10.1007/s00477-016-1216-8 , issn =

    work page doi:10.1007/s00477-016-1216-8 2017

  61. [61]

    , number =

    Esteban, Cristóbal and Arostegui, Inmaculada and Aramburu, Amaia and Moraza, Javier and Najera-Zuloaga, Josu and Aburto, Myriam and Aizpiri, Susana and Chasco, Leyre and Quintana, José M. , number =. 2020 , journal =. doi:10.1186/s12931-020-01395-z , issn =

    work page doi:10.1186/s12931-020-01395-z 2020

  62. [62]

    2023 , journal =

    Shen, Shirun and Zhou, Huiya and He, Kejun and Zhou, Lan , publisher =. 2023 , journal =. doi:10.1007/s13253-023-00585-8 , issn =

    work page doi:10.1007/s13253-023-00585-8 2023

  63. [63]

    Berrendero, J. R. and Justel, A. and Svarc, M. , number =. 2011 , journal =. doi:10.1016/J.CSDA.2011.03.011 , issn =

    work page doi:10.1016/j.csda.2011.03.011 2011

  64. [64]

    , number =

    Garcia-Rio F, Rojo B, Casitas R, Lores V, Madero R, Romero D, Galera R, Villasante C. , number =. 2012 , journal =

    work page 2012

  65. [65]

    2019 , journal =

    Descary, M-H and Panaretos, V M , number =. 2019 , journal =. doi:10.1093/biomet/asy055 , issn =

    work page doi:10.1093/biomet/asy055 2019

  66. [66]

    2000 , booktitle =

    Lee, Thomas C M , volume =. 2000 , booktitle =

    work page 2000

  67. [67]

    and Lange, P

    Garcia-Aymerich, J. and Lange, P. and Benet, M. and Schnohr, P. and Ant. 2006 , journal =. doi:10.1136/thx.2006.060145 , issn =

    work page doi:10.1136/thx.2006.060145 2006

  68. [68]

    Ruppert, David and Wand, M. P. and Carroll, Raymond J. , pages =. 2009 , journal =. doi:10.1214/09-EJS525 , issn =

    work page doi:10.1214/09-ejs525 2009

  69. [69]

    2017 , journal =

    Trevethan, Robert , month =. 2017 , journal =. doi:10.3389/fpubh.2017.00307 , issn =

    work page doi:10.3389/fpubh.2017.00307 2017

  70. [70]

    2022 , journal =

    Masak, T and Sarkar, S and Panaretos, V M , pages =. 2022 , journal =

    work page 2022

  71. [71]

    and Bharath, Karthik and Kurtek, Sebastian and Brophy, Juliet K

    Matthews, Gregory J. and Bharath, Karthik and Kurtek, Sebastian and Brophy, Juliet K. and Thiruvathukal, George K. and Harel, Ofer , month =. 2021 , journal =. doi:10.3389/fams.2021.759622 , issn =

    work page doi:10.3389/fams.2021.759622 2021

  72. [72]

    , number =

    Goldsmith, Jeff and Huang, Lei and Crainiceanu, Ciprian M. , number =. 2014 , journal =. doi:10.1080/10618600.2012.743437 , issn =

    work page doi:10.1080/10618600.2012.743437 2014

  73. [73]

    and Vicini, Andrea , number =

    Arnone, Eleonora and Sangalli, Laura M. and Vicini, Andrea , number =. 2023 , journal =. doi:10.1177/1471082X211057959 , issn =

    work page doi:10.1177/1471082x211057959 2023

  74. [74]

    2001 , journal =

    Zhou, Shanggang and Shen, Xiaotong , number =. 2001 , journal =. doi:10.1198/016214501750332820 , issn =

    work page doi:10.1198/016214501750332820 2001

  75. [75]

    2017 , journal =

    Panayi, Efstathios and Peters, Gareth W and Kyriakides, George , number =. 2017 , journal =

    work page 2017

  76. [76]

    2008 , journal =

    Parikh, Rajul and Mathai, Annie and Parikh, Shefali and Chandra Sekhar, ; G and Thomas, Ravi , pages =. 2008 , journal =

    work page 2008

  77. [77]

    and Porter, William C

    Russell, Brook T. and Porter, William C. , number =. 2022 , journal =. doi:10.1007/s10651-021-00505-4 , issn =

    work page doi:10.1007/s10651-021-00505-4 2022

  78. [78]

    and Wynne, Randolph H

    Brooks, Evan B. and Wynne, Randolph H. and Thomas, Valerie A. , number =. 2018 , journal =. doi:10.3390/rs10101502 , issn =

    work page doi:10.3390/rs10101502 2018

  79. [79]

    1999 , journal =

    Brumback, Babette A and Ruppert, David and Wand, M P , number =. 1999 , journal =. doi:10.2307/2669991 , issn =

    work page doi:10.2307/2669991 1999

  80. [80]

    and Crainiceanu, Ciprian and Zipunnikov, Vadim and Gellar, Jonathan , number =

    Johns, Jordan T. and Crainiceanu, Ciprian and Zipunnikov, Vadim and Gellar, Jonathan , number =. 2019 , journal =. doi:10.1080/10618600.2019.1604373 , issn =

    work page doi:10.1080/10618600.2019.1604373 2019

Showing first 80 references.