Artikel
Integrating different measurement instruments in a longitudinal clinical registry using a domain adaptation approach
Suche in Medline nach
Autoren
-
Maren Hackenberg
- Institute of Medical Biometry and Statistics, Faculty of Medicine and Medical Center, University of Freiburg, Freiburg, Germany
-
Michelle Pfaffenlehner
- Institute of Medical Biometry and Statistics, Faculty of Medicine and Medical Center, University of Freiburg, Freiburg, Germany
-
Max Behrens
- Institute of Medical Biometry and Statistics, Faculty of Medicine and Medical Center, University of Freiburg, Freiburg, Germany
-
Astrid Pechmann
- Department of Neuropediatrics and Muscle Disorders, Faculty of Medicine and Medical Center, University of Freiburg, Freiburg, Germany
-
Janbernd Kirschner
- Department of Neuropediatrics and Muscle Disorders, Faculty of Medicine and Medical Center, University of Freiburg, Freiburg, Germany
-
Harald Binder
- Institute of Medical Biometry and Statistics, Faculty of Medicine and Medical Center – University of Freiburg, Freiburg im Breisgau, Germany
| Veröffentlicht: | 6. September 2024 |
|---|
Gliederung
Text
Introduction: In a longitudinal clinical registry, different measurement instruments might have been used for assessing individuals at different time points. Integrating these measurements would be desirable for a more comprehensive understanding of patients’ disease trajectories, but is often further complicated by rather small numbers of individuals and time points, e.g., in rare diseases. To address this, we investigate the use of deep learning techniques for obtaining a joint latent representation, to which the items of different measurement instruments are mapped. This corresponds to domain adaptation, a well-established concept in computer science for image data. Using our proposed approach as an example, we evaluate the potential of domain adaptation in a longitudinal cohort setting, motivated by an application with different measurement instruments for assessing motor function skills in a registry of spinal muscular atrophy (SMA) patients.
Methods: We use variational autoencoders for mapping the items of each measurement instrument, at each time point where it is available, to a joint latent representation. There, we model the embedded trajectories of each instrument as solutions of a shared ordinary differential equation (ODE), where individual-specific ODE parameters are inferred from patients' baseline characteristics. The goodness of fit and complexity of the ODE solutions then allows to judge the measurement instrument mappings. We subsequently explore how alignment can be improved by incorporating corresponding penalty terms into model fitting. To systematically investigate the effect of differences between measurement instruments, we consider several scenarios based on modified SMA data, including scenarios where a mapping should be feasible in principle and scenarios where no perfect mapping is available.
Results: While the ODE fits are increasingly overshadowed by mis-alignment as the complexity of scenarios increases, some structure is still recovered, even if the availability of measurement instruments depends on a patient’s state. A reasonable mapping is feasible also in the more complex real SMA dataset.
Conclusion: The proposed approach can help to address the biometrical question of how to combine patients' measurements taken with different instruments at different times while dealing with limited numbers of individuals and time points per individual. Further, the results indicate that domain adaptation might be more generally useful in statistical modeling for longitudinal registry data.
The authors declare that they have no competing interests.
The authors declare that an ethics committee vote is not required.
References
- 1.
- Farahani A, Voghoei S, Rasheed K, Arabnia HR. A Brief Review of Domain Adaptation. In: Stahlbock R, Weiss GM, Abou-Nasr M, Yang CY, Arabnia HR, Deligiannidis L, editors. Advances in Data Science and Information Engineering. Cham: Springer International Publishing; 2021. p. 877–94. (Transactions on Computational Science and Computational Intelligence).
- 2.
- Guan H, Liu M. Domain Adaptation for Medical Image Analysis: A Survey. IEEE Transactions on Biomedical Engineering. 2022 Mar;69(3):1173–85.
- 3.
- Tzeng E, Hoffman J, Saenko K, Darrell T. Adversarial Discriminative Domain Adaptation. In: 2017 IEEE Conference on Computer Vision and Pattern Recognition (CVPR); 21-26 July 2017 Jul 21-26; Honolulu. 2017. p. 2962–71.
- 4.
- Pechmann A, König K, Bernert G, Schachtrup K, Schara U, Schorling D, et al. SMArtCARE - A platform to collect real-life outcome data of patients with spinal muscular atrophy. Orphanet Journal of Rare Diseases. 2019 Jan 21;14(1):18.
- 5.
- Hackenberg M, Pfaffenlehner M, Behrens M, Pechmann A, Kirschner J, Binder H. Investigating a domain adaptation approach for integrating different measurement instruments in a longitudinal clinical registry. arXiv. 2023. arXiv:2312.00616. DOI:
10.48550/arXiv.2312.00616
