Unsupervised domain transfer: Overcoming signal degradation in sleep monitoring by increasing scoring realism

Mohammad Ahangarkiasari; Andreas Tind Damgaard; Casper Haurum; Kaare B. Mikkelsen

cs.LGApr 15, 2026

Unsupervised domain transfer: Overcoming signal degradation in sleep monitoring by increasing scoring realism

Authors: Mohammad Ahangarkiasari, Andreas Tind Damgaard, Casper Haurum, Kaare B. Mikkelsen

Organizations: Department of Electrical and Computer Engineering, Aarhus University, Denmark

Abstract

Objective: Investigate whether hypnogram 'realism' can be used to guide an unsupervised method for handling arbitrary types of signal degradation in mobile sleep monitoring. Approach: Combining a pretrained, state-of-the-art 'u-sleep' model with a 'discriminator' network, we align features from a target domain with a feature space learned during pretraining. To test the approach, we distort the source domain with realistic signal degradations, to see how well the method can adapt to different types of degradation. We compare the performance of the resulting model with best-case models designed in a supervised manner for each type of transfer. Main Results: Depending on the type of distortion, we find that the unsupervised approach can increase Cohen's kappa with as little as 0.03 and up to 0.29, and that for all transfers, the method does not decrease performance. However, the approach never quite reaches the estimated theoretical optimal performance, and when tested on a real-life domain mismatch between two sleep studies, the benefit was insignificant. Significance: 'Discriminator-guided fine tuning' is an interesting approach to handling signal degradation for 'in the wild' sleep monitoring, with some promise. In particular, what it says about sleep data in general is interesting. However, more development will be necessary before using it 'in production'.

Figures & tables

Figure 1: Example hypnogram illustrating the progression of sleep stages across a single night. The x-axis represents time in hours, and the y-axis denotes sleep stages based on the AASM categories: W (Wake), R (REM), N1, N2, N3, and U (Unknown/non-scorable).

Dataset	Subjects	Records	Dataset	Subjects	Records
ABC	49	132	CCSHS	515	515
CFS	-	730	DCSM	255	255
DOD-H	25	25	DOD-O	55	55
EESM19	20	200	HOMEPAP	-	233
ISRUC-SG1	100	91	ISRUC-SG2	8	6
ISRUC-SG3	10	10	MASS-C1	53	40

Table 1: Datasets used in this study. Further details of each data set are found in the following references: ABC: [ 1 ] CCSHS: [ 41 ] , CFS: [ 33 ] , DOD: [ 11 ] , EESM19: [ 23 ] , HOMEPAP: [ 34 ] , DCSM: [ 41 ] , ISRUC: [ 16 ] , MASS: [ 27 ] , PHYS: [ 8 ] , SEDF : [ 15 ] , SVUH: [ 9 ] , SOF: [ 37 ] . Where possible, data was obtained through the ’National Sleep Research Resource’ [ 41 ] .

Figure 2: The overall structure of the proposed model. $E_{\text{s}}$ denotes the frozen, pretrained encoder used for the source domain, while $E_{\text{t}}$ is initialized with the same pretrained U-Sleep weights but is fine-tuned during training to adapt to the target domain. $Dec$ is the pretrained U-Sleep Decoder and $D$ is the discriminator. $f_{s}$ and $f_{t}$ denote the decoder output features of the U-Sleep model for the source and target domains, respectively (refer to section 4.3 ).

Figure 3: Histogram of Chance Kappa scores. Kappa scores are computed from random EEG-EOG sample pairs drawn from the target and source datasets. Blue: Pretrained U-Sleep on the source dataset. Red: Fine-tuned U-Sleep on the target dataset.

Panel of Figure 4 — Figure 4: Comparison of Cohen’s kappa ( $\kappa$ ) before and after fine-tuning with white noise distortion. Results are compared to both pretrained model and benchmark model.

Metric	Model	EEG	EOG
$\kappa$	Pretrained model	0.23	0.02
	Finetuned model	0.42	0.18
	$\Delta$ (Fine–Pre)	0.29	0.16
	p-value	$<1\times 10^{-4}$	$<1\times 10^{-4}$
Accuracy	Pretrained model	0.46	0.31
	Finetuned model	0.60	0.42

Table 2: Effect of finetuning to ignore a noisy input channel. The column header indicates which U-Sleep input was replaced with white noise. Results are reported for Cohen’s

\kappa

, accuracy, and F1-score. P-values test whether pretrained and fine-tuned performance might be the same.

		Frequency (Hz)
Metric	Model	20	25	30
$\kappa$	Pretrained	0.30	0.29	0.28
	Fine-tuned	0.41	0.38	0.36
	$\Delta$ (Fine–Pre)	0.11	0.09	0.08
	p-value	$<1\times 10^{-4}$	$<1\times 10^{-4}$	$<1\times 10^{-4}$
Accuracy	Pretrained	0.49	0.48	0.48

Table 3: Effect of amplifier-overload distortion frequency on performance. Results are reported for Cohen’s

\kappa

, Accuracy, and F1-score. The fine-tuned model exceeds the pretrained baseline across all frequencies.

Panel of Figure 5 — Figure 5: Cohen’s kappa ( $\kappa$ ) between fine-tuned and pretrained U-Sleep models under different low-cut thresholds (1, 5, 7, and 10 Hz) in a mid-pass filter with a fixed high-cut frequency of 20 Hz.

		Low-cut frequency (Hz)
Metric	Model	1	5	7	10
$\kappa$	Pretrained	0.68	0.36	0.33	0.27
	Fine-tuned	0.71	0.52	0.47	0.42
	$\Delta\kappa$ (Fine–Pre)	0.03	0.16	0.15	0.15
	p-value	$<1\times 10^{-4}$	$<1\times 10^{-4}$	$<1\times 10^{-4}$	$<1\times 10^{-4}$
Accuracy	Pretrained	0.77	0.55	0.53	0.45

Table 4: Effect of increasing high-pass (low-cut) frequency on performance for different metrics. The fine-tuned model outperforms the pretrained baseline across all settings. P-values test whether pretrained and fine-tuned performance might be the same.

Class	Support	Pretrained Model			pre-trained Model
Class	Support	Precision	Recall	F1-score	Precision	Recall	F1-score
W	27467	0.5593	0.5485	0.5538	0.5940	0.6161	0.6048
N1	6664	0.2412	0.2758	0.2574	0.2420	0.2938	0.2654
N2	45262	0.5674	0.6449	0.6037	0.6203	0.6547	0.6370
N3	16756	0.6886	0.2730	0.3910	0.7630	0.4530	0.5685
REM	14221	0.3508	0.4371	0.3892	0.4245	0.4796	0.4504

Table 5: Per-class performance comparison against GT labels (support = number of ground-truth samples(epochs) per class).

Figure 6: Effect of the training set size on the fine-tuned model’s Kappa under different noise conditions. The plot compares Kappa values obtained with white noise and spectral deformation and amplifier overloaded distortions as the proportion of training samples increases from 10% to 90%.

Appendix figures & tables4 assets

Supplementary material from the paper’s appendix.

Appendix

Panel of Figure 7 — Figure 7: (a)–(c) Visualizations of three representative distortions: bad filtering, amplifier overload, and additive white noise. The blue curve denotes the clean signal, and the red (dotted) curve denotes the distorted signal.

Figure 8: An example of band-pass filter applied to the signal with a low cut-off of 5.00 Hz and a high cut-off of 20.0 Hz, suppressing low-frequency drift and high-frequency noise while retaining features in the 5–20 Hz band.

Panel of Figure 9 — Figure 9: (a) Training curves of the generator and discriminator. (b) Curves of F1-score, accuracy, and Cohen’s kappa on the test set during training, for the ’spectral deformation’ distortion.

Panel of Figure 10 — Figure 10: Comparison of pairwise kappa scores among the pre-trained, fine-tuned, and supervisedly trained U-Sleep models. The diagonal red line represents the line of equality ( $y=x$ ). (a) illustrates the comparison between the pre-trained and fine-tuned U-Sleep models, and (b) shows the comparison between the fine-tuned and supervisedly trained U-Sleep models.

Unsupervised Domain Adaptation

Unsupervised domain adaptation transfers models from labeled source data to unlabeled target domains with different data distributions.

See all 183 papers

Pretraining on Sleep Data Improves non-Sleep Biosignal Tasks
May 4, 2026William Lehn-Schiøler, Magnus Ruud Kjær, Phillip Hempel +7Self-Supervised Time Series Representation Learning Cross-Domain Transfer Learning
BrainCapture ApS, Kongens Lyngby, Denmark · DTU Health Tech, Technical University of Denmark, Kongens Lyngby, Denmark · DTU Compute, Technical University of Denmark, Kongens Lyngby, Denmark +4
DualStabSleepNet: A Dual-Domain Diffusion Stabilization Network for Robust Sleep Staging
Aug 17, 2026Chongjian Wang, Chen Liu, Junjie Gao +3Diffusion-Based Denoising Sleep Stage Classification
School of Mathematics and Systems Science, Shandong University of Science and Technology, Qingdao, 266590, Shandong, China · School of Artificial Intelligence, Shandong Women’s University, Jinan, 250352, Shandong, China · School of Computer Science and Engineering, South China University of Technology, Guangzhou, 510641, Guangdong, China
Towards Multi-Source Domain Generalization for Sleep Staging with Noisy Labels
Apr 11, 2026Kening Wang, Di Wen, Yufan Chen +6Sleep Stage Classification Domain Generalization
Karlsruhe Institute of Technology · ETH Zurich · Hunan University +1

May 4, 2026J/K move · Enter open · S save

Pretraining on Sleep Data Improves non-Sleep Biosignal Tasks

William Lehn-Schiøler, Magnus Ruud Kjær, Phillip Hempel, Magnus Guldberg Pedersen+6

BrainCapture ApS, Kongens Lyngby, Denmark · DTU Health Tech, Technical University of Denmark, Kongens Lyngby, Denmark · DTU Compute, Technical University of Denmark, Kongens Lyngby, Denmark+4

Self-Supervised Time Series Representation Learning Cross-Domain Transfer Learning

Open paper AskPDF arXiv

Sleep foundation models have recently demonstrated strong performance on in-domain polysomnography tasks, including sleep staging, apnea detection, and disease risk prediction. In this work, we investigate whether sleep biosignals can serve as an effective pretraining distribution for learning representations that transfer beyond sleep to adjacent domains. Following sleep foundation models, we perform sleep-only multimodal contrastive pretraining (with a leave-one-out objective) and evaluate transfer to non-sleep EEG and ECG, two well-benchmarked biosignal modalities with heterogeneous datasets and clinically meaningful downstream tasks. Across eight downstream tasks spanning multiple EEG and ECG datasets, sleep pretraining consistently improves performance relative to training from scratch. Moreover, on several tasks, we achieve performance competitive with or surpassing prior specialized state-of-the-art and foundation models.

Abstract

Unsupervised Domain Adaptation

Explore similar work

Pretraining on Sleep Data Improves non-Sleep Biosignal Tasks

DualStabSleepNet: A Dual-Domain Diffusion Stabilization Network for Robust Sleep Staging

Towards Multi-Source Domain Generalization for Sleep Staging with Noisy Labels