EUVIMEDExplore EUVIMED
Open access · CC BY 4.0

Using Latent Representations to Link Disjoint Longitudinal Data for Mixed‐Effects Regression

Clemens Schächter iD, Maren Hackenberg iD, Michelle Pfaffenlehner iD, Félix B Tambe-Ndonfack iD, Thorsten Schmidt iD, Astrid Pechmann, Janbernd Kirschner iD, Jan Hasenauer iD, Harald Binder iD

Statistics in Medicine · 2026-08 · 45 (18-19)

DOI: 10.1002/sim.70701 ↗

At a glance

Journal article in Statistics in Medicine, published online on 2026-08-20. Written by Clemens Schächter and 8 co-authors. Institution: Institute of Medical Biometry and Statistics (IMBI), Medical Faculty and Medical Center, University of Freiburg, Freiburg, Germany. Major MEDLINE topics: Models, Statistical. 66 references. The full text is freely available (open access, CC BY 4.0). Funded by: Deutsche Forschungsgemeinschaft.

AbstractSubjectsBibliographic detailsCiteRelated studies

Abstract (original language)

Many rare diseases offer limited established treatment options, leading patients to switch therapies when new medications emerge. To analyze the impact of such treatment switches within the low sample size limitations of rare disease trials, it is important to use all available data sources. This, however, is complicated when the use of measurement instruments changes during the observation period, for example when instruments are adapted to specific age ranges. The resulting disjoint longitudinal data trajectories complicate the application of traditional modeling approaches like mixed‐effects regression. We tackle this by mapping observations of each instrument to an aligned low‐dimensional temporal trajectory, enabling longitudinal modeling across instruments. Specifically, we employ a set of variational autoencoder architectures to embed item values into a shared latent space for each time point. Temporal disease dynamics and treatment switch effects are then captured through a mixed‐effects regression model applied to latent representations. To enable statistical inference, we present a novel statistical testing approach that accounts for the joint parameter estimation of mixed‐effects regression and variational autoencoders. The methodology is applied to quantify the impact of treatment switches for patients with spinal muscular atrophy. Here, our approach aligns motor performance items from different measurement instruments for mixed‐effects regression and maps estimated effects back to the observed item level to quantify the treatment switch effect. Our approach allows for model selection as well as for assessing effects of treatment switching. The results highlight the potential of modeling in joint latent representations for addressing small data challenges.

Subjects

MeSH terms (MEDLINE)

Models, StatisticalHumansLongitudinal StudiesRegression AnalysisAutoencoder

Author keywords

measurement instrument alignmentmixed‐effects regressionsmall dataspinal muscular atrophystatistical inferencetreatment switchesvariational autoencoder

Authors and institutions

  • Institute of Medical Biometry and Statistics (IMBI), Medical Faculty and Medical Center, University of Freiburg, Freiburg, Germany.
  • Freiburg Center for Data Analysis, Modeling and AI, University of Freiburg, Freiburg, Germany.
  • Department of Mathematical Stochastics, Mathematical Institute, University of Freiburg, Freiburg, Germany.
  • Department of Neuropediatrics and Muscle Disorders, Medical Faculty and Medical Center, University of Freiburg, Freiburg, Germany.
  • Bonn Center for Mathematical Life Sciences, University of Bonn, Bonn, Germany.
  • Life and Medical Sciences (LIMES) Institute, University of Bonn, Bonn, Germany.

ORCID profiles: Clemens Schächter, Maren Hackenberg, Michelle Pfaffenlehner, Félix B Tambe-Ndonfack, Thorsten Schmidt, Janbernd Kirschner, Jan Hasenauer, Harald Binder

Funding

  • Deutsche Forschungsgemeinschaft

Cite

APA

Schächter, C., Hackenberg, M., Pfaffenlehner, M., Tambe-Ndonfack, F. B., Schmidt, T., Pechmann, A., Kirschner, J., Hasenauer, J., & Binder, H. (2026). Using Latent Representations to Link Disjoint Longitudinal Data for Mixed‐Effects Regression. Statistics in Medicine, 45(18-19). https://doi.org/10.1002/sim.70701

Vancouver

Schächter C, Hackenberg M, Pfaffenlehner M, Tambe-Ndonfack FB, Schmidt T, Pechmann A, et al. Using Latent Representations to Link Disjoint Longitudinal Data for Mixed‐Effects Regression. Statistics in Medicine. 2026;45(18-19). doi:10.1002/sim.70701

BibTeX
@article{Schchter2026,
  author = {Schächter, Clemens and Hackenberg, Maren and Pfaffenlehner, Michelle and Tambe-Ndonfack, Félix B and Schmidt, Thorsten and Pechmann, Astrid and Kirschner, Janbernd and Hasenauer, Jan and Binder, Harald},
  title = {Using Latent Representations to Link Disjoint Longitudinal Data for Mixed‐Effects Regression},
  journal = {Statistics in Medicine},
  year = {2026},
  volume = {45},
  number = {18-19},
  doi = {10.1002/sim.70701}
}
Start for free →