DyssomniaNN: Multi-Scale Spiking Neural Modeling of Human Sleep Disorders and Perceived Restedness

Daniel Szelogowski

cs.NEOct 9, 2026

DyssomniaNN: Multi-Scale Spiking Neural Modeling of Human Sleep Disorders and Perceived Restedness

Authors: Daniel Szelogowski

Organizations: Department of Computer Science University of Wisconsin–Whitewater

Abstract

Clinical sleep medicine repeatedly documents profound divergences between objective sleep duration and subjective daytime impairment, notably exemplified by sleep state misperception in paradoxical insomnia and boundary instability in narcolepsy type 1. Here, we present DyssomniaNN: a multi-scale biophysical computational framework coupling Borbély's two-process regulation of sleep homeostasis and circadian drive with Saper's hypothalamic reciprocal flip-flop circuit and a recurrent corticothalamic leaky integrate-and-fire spiking neural network. We formalize five clinical cohorts across an identical 72-hour photoperiod: healthy control, circadian rhythm sleep-wake disorder (delayed-phase), chronic insomnia, paradoxical insomnia, and narcolepsy type 1. The synthetic local field potential converges closely with empirical human electroencephalography spectra from the PhysioNet Sleep-EDF database across all canonical frequency bands (spectral mean absolute error of 0.33--0.35). Furthermore, we introduce the Well-Rested Index to quantify restorative efficacy. Simulations reveal that sleep duration fails to predict restorative recovery: paradoxical insomnia achieves normal total sleep time (21.15 h) yet yields low restoration (37.6%) from unsuppressed cortical beta-band intrusion, whereas hypocretin deficiency in narcolepsy produces severe nocturnal fragmentation (781 min of wake after sleep onset) and 35 sleep-onset rapid eye movement periods. DyssomniaNN provides an in silico platform for mechanistic sleep research and virtual neuropharmacology.

Figures & tables

Panel of Figure 1 — Figure 1: Architectural topology of the DyssomniaNN framework. Borbély’s two-process regulators ( $S(t)$ and $C(t)$ ) drive the reciprocal hypothalamic flip-flop switch (VLPO $\leftrightarrow$ LHA), where $+w_{sv}S(t)$ represents homeostatic somnogenic excitation of the VLPO and $+w_{ca}C(t)$ represents circadian alerting excitation of the LHA. The resulting subcortical arousal level gates the corticothalamic LIF network, whose synthetic LFP is evaluated against empirical PSG spectra from the PhysioNet Sleep-EDF database.

Biophysical Parameter	Normal	CRSWD	Chronic Insomnia	Paradoxical	Narcolepsy
Phase Offset $\Delta\phi$ (h)	$0.0$	$\mathbf{+4.0}$	$0.0$	$0.0$	$0.0$
Circadian Amplitude $A$	$0.38$	$\mathbf{0.20}$	$0.38$	$0.38$	$0.38$
Orexin Tone $\Omega_{\text{tone}}$	$1.0$	$1.0$	$1.0$	$1.0$	$\mathbf{0.05}$
Arousal Bias $I_{\text{arousal}}$ (pA)	$0.0$	$0.0$	$\mathbf{+0.60}$	$0.0$	$-0.10$
VLPO Inhibitory Weight $w_{av}$	$2.1$	$2.1$	$\mathbf{0.95}$	$2.1$	$2.1$
Cortical Inh. Scale $\gamma_{\text{inh}}$	$1.0$	$1.0$	$1.0$	$\mathbf{0.25}$	$1.0$

Table 1: Biophysical parameterization across the five simulated clinical cohorts. Values departing from normal control physiology are bolded and grounded in empirical clinical findings [ 31 , 38 , 30 , 20 , 25 , 34 , 11 ] .

Cohort	Sleep%	TST	SOL	WASO	Cycles	SOREMPs	Rested%	Rest Status	Spectral MAE
	(Env)	(h)	(min)	(min)					(vs PhysioNet)
Normal Control	$29.4\%$	$21.15$	$29$	$172$	$13$	$0$	$\mathbf{86.4\%}$	Optimal	$0.3369$
CRSWD (Delayed)	$23.1\%$	$16.63$	$99$	$454$	$11$	$0$	$\mathbf{47.9\%}$	Phase-Shifted	$0.3426$
Chronic Insomnia	$16.8\%$	$12.10$	$194$	$525$	$8$	$0$	$\mathbf{43.3\%}$	Fragmented	$0.3449$
Paradoxical Insomnia	$29.4\%$	$21.15$	$29$	$172$	$13$	$0$	$\mathbf{37.6\%}$	Non-Restorative	$0.3449$
Narcolepsy Type 1	$36.4\%$	$26.18$	$29$	$781$	$17$	$35$	$\mathbf{48.2\%}$	Fragmented	$0.3349$

Table 2: Clinical PSG Metrics and Restorative Telemetry over 72.0 Simulated Hours. Definitions: Sleep% is total sleep duration divided by observation window; TST is Total Sleep Time; SOL is Sleep Onset Latency; WASO is Wake After Sleep Onset; Cycles is completed ultradian NREM-REM cycles; SOREMPs is Sleep-Onset REM Periods (

\leq 15\,\text{min}

); Rested% is the Well-Rested Index (

\mathcal{R}

); Spectral MAE is relative power error versus PhysioNet Sleep-EDF C3-A2 N3 baseline.

Figure 2: Continuous 72-hour multi-day simulation across the five clinical cohorts. (A) Homeostatic sleep pressure $S(t)$ and circadian pacemaker $C(t)$ trajectories against environmental photoperiods (shaded regions denote nocturnal intervals, 23:00 to 07:00). Homeostatic trajectories $S(t)$ for paradoxical insomnia (dashed purple) identically coincide with normal controls (solid blue) due to intact subcortical homeostatic kinetics. Note the $+4.0\,\text{h}$ phase delay in CRSWD and daytime somnogenic clearances in narcolepsy. (B) Concurrent PSG hypnograms scored via AASM criteria (W=Wake, N1–N3=NREM stages, R=REM), displaying physiological 90-minute ultradian cycling in normal sleep, prolonged nocturnal awakenings in chronic insomnia, and daytime sleep attacks in narcolepsy.

Figure 3: Subcortical Saper reciprocal flip-flop switch dynamics. (A) Sleep-active VLPO GABAergic firing rate $V(t)$ . (B) Wake-active LHA orexin/monoaminergic firing rate $A(t)$ . The normal control displays sharp, bistable transitions with pronounced hysteresis. In contrast, narcolepsy type 1 exhibits crippled LHA tone ( $\Omega_{\text{tone}}=0.05$ ), inducing state instability, nocturnal sleep fragmentation, and involuntary daytime transitions.

Figure 4: Mechanistic resolution of the sleep state misperception observed in paradoxical insomnia. (A) Objective PSG hypnograms for healthy control and paradoxical insomnia, illustrating identical macroscale architecture and total sleep duration ( $21.15\,\text{h}$ ). (B) Cortical beta band ( $15.0$ – $30.0\,\text{Hz}$ ) relative power; deficient cortical GABAergic interneuron inhibition ( $\gamma_{\text{inh}}=0.25$ ) permits pathological high-frequency beta power intrusion throughout stage N3 slow-wave sleep. (C) Subjective wakefulness meter, showing that persistent cortical hyperarousal induces a perceived wakefulness state of $85\%$ – $100\%$ despite objective electrophysiological sleep.

Figure 5: Empirical validation against human PSG from the PhysioNet Sleep-EDF Database ( SC4001E0-PSG.edf ). (A) Relative spectral band power distribution during stage N3 slow-wave sleep comparing empirical human EEG (channel C3-A2) against the synthetic LFP generated by the recurrent LIF network. (B) Continuous 72-hour Mean Absolute Error (MAE) trajectories across the five clinical cohorts, converging within $\text{MAE}\approx 0.33$ – $0.35$ .

Appendix figures & tables2 assets

Supplementary material from the paper’s appendix.

Appendix

Abbr.	Full Term	Domain/Description
AASM	American Academy of Sleep Medicine	Clinical scoring standard organization
AMPA	$\alpha$ -amino-3-hydroxy-5-methyl-4-isoxazolepropionic acid	Fast excitatory ionotropic glutamate receptor
ATP	Adenosine Triphosphate	Cellular metabolic energy currency
CRSWD	Circadian Rhythm Sleep-Wake Disorder	Intrinsic circadian timing pathology
CSF	Cerebrospinal Fluid	Clear fluid surrounding brain and spinal cord
DLMO	Dim Light Melatonin Onset	Gold-standard circadian phase biomarker

Appendix

Table 3: Comprehensive summary of medical, physiological, and computational abbreviations.

Chemical	Class & Receptors	Anatomical Projections	Primary Functional Role
GABA	Amino Acid $\text{GABA}_{A}$ (ionotropic $\text{Cl}^{-}$ ) $\text{GABA}_{B}$ (metabotropic $G_{i}$ )	VLPO, basal forebrain, local cortical interneurons	Primary central nervous system inhibitory neurotransmitter. Hyperpolarizes target neurons, gates slow-wave sync, and inhibits ascending monoamines to initiate sleep.
Galanin	Neuropeptide GalR1, GalR2, GalR3 ( $G_{i}/G_{q}$ )	Co-localized with GABA in VLPO neurons	Co-released with GABA; provides sustained slow inhibition onto orexinergic and monoaminergic wake systems to maintain sleep continuity.
Orexin/ Hypocretin	Neuropeptide OX1R ( $G_{q}$ ), OX2R ( $G_{q}/G_{i}$ )	Lateral hypothalamus (LHA) to brainstem, cortex, and thalamus	Essential wakefulness stabilizer. Projects excitatory drive to monoaminergic nuclei, reinforcing flip-flop hysteresis and preventing cataplexy/SOREMPs.
Glutamate	Amino Acid AMPA, NMDA, Kainate, mGluRs	Cortical pyramidal cells, thalamic relay, parabrachial nucleus	Primary fast excitatory neurotransmitter. Mediates cognitive processing, intracortical communication, and ascending reticular arousal drive.
Acetylcholine (ACh)	Quaternary Amine Muscarinic ( $M_{1}$ – $M_{5}$ ), Nicotinic	Basal forebrain, PPT/LDT brainstem nuclei	Desynchronizes cortical EEG. Active during waking cognition and elevated during REM sleep; quiescent during deep N3 slow-wave sleep.
Noradrenaline/ Norepinephrine	Monoamine $\alpha_{1},\alpha_{2},\beta_{1},\beta_{2}$ adrenergic	Locus coeruleus (LC) to neocortex, thalamus, spinal cord	Mediates vigilance, environmental alertness, and stress responses. Discharges rapidly during wakefulness; quiescent during NREM and silent during REM sleep.

Appendix

Table 4: Reference guide to key neurotransmitters and neuromodulators in sleep-wake neurobiology.

Explore similar work

Dec 16, 2025cs.LG

AnySleep: a channel-agnostic deep learning system for high-resolution sleep staging in multi-center cohorts

Sleep is essential for health, yet studying its dynamics requires manual sleep staging, a labor-intensive step in research and clinical care. Across centers, polysomnography (PSG) recordings are traditionally scored in 30-s epochs for pragmatic, not physiological, reasons and vary in electrode count, montage, and subject characteristics. These constraints challenge harmonized multi-center studies and the discovery of robust biomarkers on shorter timescales. We present AnySleep, a deep neural network that scores sleep from any electroencephalography (EEG) or electrooculography (EOG) data at adjustable temporal resolutions. We trained and validated the model on over 20,000 overnight recordings (> 200,000 hours of EEG and EOG) from 28 datasets across multiple clinics to promote robust generalization across sites. The model attains state-of-the-art performance and surpasses or equals established baselines at 30-s epochs. Performance improves with more channels, yet remains strong when EOG is absent or only EOG or single EEG derivations (frontal, central, or occipital) are available. On sub-30-s timescales, the model captures short wake intrusions consistent with arousals and improves prediction of pathophysiological conditions (obstructive sleep apnea, narcolepsy type 1, insomnia) over 30-s scoring. We make the model publicly available to facilitate large-scale studies with heterogeneous electrode setups and accelerate biomarker discovery in sleep.

Niklas Grieger, Jannik Raskob, Siamak Mehrkanoon +1

Department of Medical Engineering and Technomathematics, FH Aachen University of Applied Sciences, 52428 Jülich, Germany · Department of Information and Computing Sciences, Utrecht University, Utrecht, The Netherlands · Institute for Data-Driven Technologies, FH Aachen University of Applied Sciences, 52428 Jülich, Germany

Electroencephalography Sleep Stage Classification EEG Decoding Domain Generalization

Ask PDF

Sep 12, 2026cs.LG

Exploring napping paradigm for Recurrent Spiking Neural Networks

Biological organisms minimize free energy by balancing two competing demands on their internal world model: it must be accurate enough to predict sensory input, yet simple enough to generalize beyond it. Two mechanisms regulate this balance offline: sleep reduces complexity through gradual synaptic downscaling, while stochastic noise attenuates precision, relaxing the constraint sensory input imposes on synaptic reorganization. Engineered Spiking Neural Networks (SNNs) leave this balance unaddressed, favoring instantaneous, noiseless weight normalization instead. This paper investigates the hypothesis that a biologically inspired micro-sleep paradigm, napping -- combining proportional weight scaling with continuous stochastic membrane activity -- can replicate the stability of normalization while shedding model complexity. We evaluate this in an unsupervised recurrent SNN trained via trace-based spike-timing-dependent plasticity (STDP) on Gabor-preprocessed MNIST. We tune napping across three regularization regimes by sweeping its duration and membrane noise level, then compare the best configuration against weight normalization. Across all three regimes, well-tuned napping matches the accuracy of normalization: accuracy peaks at brief durations and low noise, then declines monotonically as either grows. Clustering diverges, with the strongest geometric separation arising at longer durations and higher noise -- the two terms of free energy pulling apart, accuracy rewarding data fit and structure rewarding the simpler representation that gradual, noisy downscaling induces. This gain carries a simulation cost normalization avoids, so napping is most compelling where representational structure, rather than raw classification efficiency, is the priority.

Andreas Massey, Stefano Nichele, Aliaksandr Hubin

The Norwegian University of Life Sciences, Elizabeth Stephansens v. 15, 1433 Ås, Norway · Østfold University of Applied Sciences, B R A Veien 4, 1757 Halden, Norway

Spiking Neural Networks Recurrent Neural Networks

Ask PDF

Jul 4, 2026cs.LG

Omni-Sleep: A Sleep Foundation Model via Hierarchical Contrastive Learning of CNS-ANS Dynamics

Sleep physiology arises from the coordinated dynamics of the central nervous system (CNS) and autonomic nervous system (ANS), as reflected by multimodal polysomnography signals including EEG, EOG, EMG, ECG, and respiration. However, existing sleep foundation models often fuse heterogeneous biosignals in a topology-agnostic manner, overlooking their physiological organization. We introduce Omni-Sleep, a sleep foundation model that uses the CNS/ANS partition as a physiological prior for topology-constrained representation learning. Omni-Sleep learns structured representations through three objectives: intra-system consistency, which captures shared subsystem-level factors within neural and cardio-respiratory signals; inter-system synchronization, which aligns subsystem trajectories to model brain--body dynamics; and latent-space masked temporal modeling, which captures long-horizon sleep dynamics. Pre-trained on over 100,000 hours of multi-center multimodal PSG data, Omni-Sleep is evaluated on sleep staging and multi-disease classification. Across datasets and modality-ablation settings, Omni-Sleep outperforms strong foundation-model baselines, showing improved label efficiency, cross-dataset generalization, and robustness to missing modalities. These results highlight the value of physiological hierarchy for generalizable sleep representation learning. Code is available at https://github.com/AutoBrain-sleep/OmniSleep.

Zhoujie Hou, Song Wang, Kexin Lou +3

Department of Biomedical Engineering, Southern University of Science and Technology, Shenzhen, China · Omni-Intelligence, Shenzhen, China · Shenzhen Loop Area Institute, Shenzhen, China

Sleep Stage Classification Latent Dynamics Modeling Hierarchical Representation Learning Multimodal Contrastive Learning Time Series Representation Learning

Ask PDF