{"paper_id":"3cc5aa7c-4ed6-47f6-8eb6-ffff5cf00ee7","body_text":"High-Density Electroencephalographic Analysis of Neuropathic Pain - A Prospective, Multicenter Clinical Study | Research Square window.SnipcartSettings = { analytics: { enabled: false } }; (function() { var accessVector = localStorage.getItem('access_vector') || ''; window.dataLayer = window.dataLayer || []; if (accessVector) { window.dataLayer.push({ user: { profile: { profileInfo: { snid: accessVector } } } }); } })(); (function(w,d,s,l,i){w[l]=w[l]||[];w[l].push({'gtm.start':new Date().getTime(),event:'gtm.js'});var f=d.getElementsByTagName(s)[0],j=d.createElement(s),dl=l!='dataLayer'?'&l='+l:'';j.async=true;j.src='https://www.googletagmanager.com/gtm.js?id='+i+dl;f.parentNode.insertBefore(j,f);})(window,document,'script','dataLayer','GTM-K279D39R'); Browse Preprints In Review Journals COVID-19 Preprints AJE Video Bytes Research Tools Research Promotion AJE Professional Editing AJE Rubriq About Preprint Platform In Review Editorial Policies Our Team Advisory Board Help Center Sign In Submit a Preprint Cite Share Download PDF Article High-Density Electroencephalographic Analysis of Neuropathic Pain - A Prospective, Multicenter Clinical Study Ke Ma, Yun Ji, Zhiyuan Dang, Huichun Luo, Tao Li, Yongyong Ren, and 6 more This is a preprint; it has not been peer reviewed by a journal. https://doi.org/ 10.21203/rs.3.rs-8008823/v1 This work is licensed under a CC BY 4.0 License Status: Under Review Version 1 posted You are reading this latest preprint version Abstract The absence of objective, biologically-based diagnostics for neuropathic pain (NP) is a fundamental barrier to precision medicine in neurology and psychiatry. Reliance on subjective self-report leads to misdiagnosis, heterogeneous patient cohorts in clinical trials, and an inability to objectively evaluate therapeutic efficacy. To address this, we leveraged the China Chronic Pain Cohort (CPCC) and high-density electroencephalography (HD-EEG) to identify objective neural signatures of NP. Our study enrolled 286 participants, including 117 NP patients, 56 non-neuropathic pain patients, and 113 healthy controls. HD-EEG analysis revealed distinct spatial functional connectivity patterns and neuronal oscillation coupling mechanisms characteristic of different pain categories. Multimodal analysis of HD-EEG revealed a distinct and reproducible neurophysiological signature of NP, characterized by aberrant oscillatory power and functional connectivity within a distributed network encompassing prefrontal, cingulate, and insular regions. Critically, we developed a novel Graph-Generative Network (GGN) model that translated these complex neural patterns into a clinically actionable tool. Our model differentiated NP from other pain types and healthy controls with exceptional accuracy (98%), and reliably predicted subjective pain intensity (VAS/NRS scores), effectively decoding a subjective state into an objective metric. These findings suggest that HD-EEG signatures can serve as objective biomarkers for NP, offering new avenues for objective pain assessment and personalized treatment strategies. Future research should focus on validating these biomarkers in larger cohorts and exploring their potential for broader clinical application. Biological sciences/Neuroscience/Peripheral nervous system/Somatic system Health sciences/Biomarkers/Predictive markers Figures Figure 1 Figure 2 Figure 3 Figure 4 Figure 5 Figure 6 Introduction A critical challenge facing pain medicine is the severe lack of objective tools for assessing pain severity and monitoring treatment efficacy [ 1 ] . Despite the advancements in drug therapies, minimally invasive interventions, and multidisciplinary collaborations aimed at tackling chronic pain, research and surveys have shown a troubling trend of high treatment dissatisfaction rates among chronic pain patients [ 2 ] . Based on the latest research and surveys, the dissatisfaction rate of chronic pain patients with analgesic effects is as high as 64%, and the rate of complete relief after treatment for chronic pain patients is less than 20% [ 3 ] . This persistent therapeutic gap is largely attributable to the reliance on subjective patient-reported outcomes like the Visual Analog Scale (VAS) and Numeric Rating Scale (NRS), which fail to provide consistent, quantitative, and biomarker-independent measures of pain experience [ 4 ] . Furthermore, existing clinical approaches often narrowly target the sensory dimension of pain, overlooking its complex emotional, cognitive, and behavioral aspects [ 5 ] .The brain serves as the foundation of human cognition, emotion, and ultimately, pain perception, which involves intricate neural networks. A growing body of evidence indicates that chronic pain is linked to extensive and measurable changes in brain structure and function [ 6 ] . Understanding these neurophysiological signatures is crucial not only for deciphering the pathophysiology of pain but also for developing the long-sought objective biomarkers essential for precise diagnosis and effective treatment. Electroencephalography (EEG), which non-invasively records the brain's electrophysiological activity, has long served as a valuable tool in pain research [ 7 ] . However, conventional EEG methodologies have predominantly relied on stimulus-evoked paradigms—typically using acute, experimentally induced pain—which suffer from critical limitations in the context of chronic pain. These approaches are not only clinically impractical due to requirements for standardized stimulation and patient cooperation, but also fail to capture the persistent, resting-state brain activity that characterizes chronic neuropathic conditions [ 7 , 8 ] . Furthermore, traditional low-density EEG systems, often limited to 32 or fewer electrodes, provide insufficient spatial sampling to resolve the distributed network dynamics underlying pain processing. The brain's complex neural architecture, involving intricate signal transmissions across widely distributed regions such as the prefrontal cortex, anterior cingulate, insula, and somatosensory areas, necessitates high-density spatial sampling to accurately map functional interactions within the pain neuromatrix. This fundamental limitation in spatial resolution has likely contributed to inconsistencies across previous studies and hindered the identification of reliable, clinically applicable neural signatures for chronic neuropathic pain. This study addresses these critical limitations. Leveraging high-density EEG (HD-EEG), which employs a significantly greater number of densely positioned electrodes, enables a far more comprehensive recording of the brain's electrical signals and network interactions. Crucially, our approach focuses on resting-state brain activity, eliminating the need for artificial pain provocation and enhancing clinical feasibility. Most significantly, we integrate these rich HD-EEG datasets with advanced computational approaches, particularly Graph Neural Network (GNN)-based models like our novel Graph-Generative Network (GGN). These powerful algorithms are designed to learn the complex spatial and temporal patterns inherent in the EEG data. This enables not only the highly accurate classification of pain types (e.g., neuropathic vs. non-neuropathic) directly from resting-state signals but also the precise prediction of subjective pain intensity levels, offering a transformative pathway towards objective, quantitative pain assessment. Conducted within the China Chronic Pain Cohort (CPCC), the largest prospective, multicenter clinical cohort of its kind in China to date, this research provides robust evidence derived from a uniquely representative sample. By observing these detailed resting-state activity patterns across the brain in this extensive cohort, researchers can more accurately identify and comprehend the neural pathways associated with chronic pain perception and its modulation, paving the way for improved management strategies. Results 3.1 Experimental design and analytical workflow of the HD-EEG study. Our high-density electroencephalography (HD-EEG) analysis of the China Chronic Pain Cohort identified robust and replicable neurophysiological signatures that not only distinguish chronic pain patients from healthy controls but also permit precise subtyping of pain conditions. As illustrated in F ig. 1 , we analyzed resting-state neural activity from 354 participants across both eyes-open and eyes-closed conditions, thereby capturing state-dependent cortical dynamics relevant to sensory processing and internal cognitive states. The analytical pipeline was structured as a hierarchically integrated computational workflow: we first conducted source reconstruction using the Brainstorm toolbox to localize group-level differences in cortical activity and examine their relationship with neuropathic pain; we then quantified spectral power and functional connectivity—including weighted phase lag index (wPLI) and phase-amplitude coupling—to characterize oscillatory profiles and network-level interactions; finally, we introduced a novel Graph-Generative Network (GGN) that integrates these spatiotemporal features to extract pain-specific biomarkers and objectively predict clinical pain intensity directly from neural activity, achieving state-of-the-art performance in both diagnostic classification and pain scoring. 3.2 The average pow er spectra across all electrodes Spectral analysis revealed fundamental oscillatory abnormalities in NP patients, forming the basis for subsequent spatial and network-level investigations. As shown in Fig. 2a, the averaged power spectral density (PSD) curves exhibited notable group differences, with NP patients demonstrating significant power enhancements in the theta (4–8 Hz) and low-alpha (8–10 Hz) bands. Statistical analysis using one-way ANOVA with FDR correction confirmed these observations across six frequency bands (Fig. 2b), revealing significant group differences in theta (F(2,241) = 7.874, p = 0.000487, η² = 0.0613) and low-alpha bands (F(2,241) = 6.302, p = 0.00215, η² = 0.0497). Critically, the peak alpha frequency (PAF) was significantly slowed in NP patients compared to both n-NP patients and healthy controls (HC) (FDR = 0.0000598, Cohen's d = -0.604) (Fig. 2c, d), aligning with the thalamocortical dysrhythmia theory. It is worth noting that the relationship of PAF with pain sensitivity and chronic pain has been pointed out in several studies, such as the inability of acute irritant pain to induce changes in PAF in subjects [9] , while in chronic pain, PAF can be used as a biomarker of individual pain sensitivity, and slower PAF can predict higher pain levels [10] . In our study, both types of chronic pain resulted in the slowest PAF, and neuropathic pain, which was more painful, had the slowest PAF, which was associated with increased pain, this shows the potential of PAF indicators as biomarkers in chronic pain. 3.3 Spectral related spatial distribution Building upon these spectral abnormalities, we mapped their spatial distribution to pinpoint the specific brain regions involved. The topographic maps in Fig. 3a illustrate the group-level spatial distribution of mean power, revealing enhanced theta and low-alpha power in NP patients, with a notable predominance in posterior regions. Fig. 3b displays pairwise PSD differences between groups, quantified by Tukey's Q-statistic, highlighting significant enhancements in NP patients compared to HC across most brain regions. The statistical distribution in Fig. 3c confirms that in the theta band, NP patients exhibited significantly higher power than HC, with the most pronounced enhancements in the occipital lobe and posterior aspects of the temporal lobes (p < 0.001). Electrode-level analysis identified the most significant enhancements at P7 (FDR = 0.0000166), O1 (FDR = 0.0000214), and O2 (FDR = 0.0000464), all located at the occipital-parietal junction. In the low-alpha band, NP patients showed significantly higher power than HC across the entire occipital lobe, as well as parts of the prefrontal and temporal lobes (p < 0.05), with maximal differences at Oz (p < 0.001). High-Density Electroencephalographic Analysis of Neuropathic Pain - A Prospective, Multicenter Clinical Study 3.4 Machine-Learning classifier Based on spatial spectral differences Leveraging these spatially distinct spectral features, we developed a LASSO logistic regression model to evaluate their collective power for objective pain classification. Our feature space comprised 762 spectral features (127 electrodes × 6 frequency bands) from 3,125 EEG segments. Fig. 4a presents the band average power for the top three electrodes with the highest SHAP values, underscoring the importance of specific electrodes in the theta and low-alpha bands for classification. The LASSO model with optimal regularization (λ = 0.05) retained 107 features (14.4% of original features), achieving an overall accuracy of 83% in the three-class classification task (Fig. 4b), with particularly high performance in identifying NP samples (accuracy = 0.85, recall = 0.88). Fig. 4c illustrates the SHAP value feature importance, confirming that the theta and low-alpha bands were the most critical for classification (mean SHAP: 0.518 and 0.571, respectively), consistent with our univariate analyses. The model demonstrated AUC values of 0.685 for NP versus other categories, 0.605 for n-NP, and 0.723 for HC. 3.5 Brain area coherence and functional connectivity To investigate whether NP involves large-scale network dysfunction beyond local spectral changes, we analyzed functional connectivity and cross-frequency interactions using the weighted phase lag index (wPLI) and phase-amplitude coupling (PAC). Fig. 5a shows the average wPLI connectivity strength across different frequency bands for the three groups, indicating enhanced theta and low-alpha band connectivity in NP patients. wPLI analysis demonstrated widespread enhancements in functional connectivity for NP patients, most prominently in the theta band (336 connections significantly enhanced vs. HC, Bonferroni-corrected p < 0.01). Fig. 5b and 5c visualize the significant connectivity differences in the theta band between NP patients and HC, and between n-NP patients and HC, respectively, with key strengthened pathways involving prefrontal, cingulate, and insular regions. Notable enhanced connections included those between the left caudal anterior cingulate cortex and right precuneus (p = 0.000011), and between the right isthmus cingulate cortex and left rostral middle frontal gyrus (p = 0.0000098). Phase-amplitude coupling (PAC) analysis further revealed enhanced theta-gamma coupling between occipital and frontal regions in NP (Fig. 5d), suggesting altered cross-frequency communication underlying pain processing. 3.6 Graph Generative Network (GGN) Finally, to integrate these local and network-level features into a unified, clinically translatable model, we developed a Graph-Generative Network (GGN) to dynamically model brain connectivity from HD-EEG signals. The GGN architecture is depicted in Fig. 6a, featuring a novel connectivity graph generator that learns dynamic functional connectivity patterns from 127-channel EEG data. The model achieved exceptional performance, differentiating NP, n-NP, and HC with an overall accuracy of 98%, substantially outperforming the LASSO classifier (Fig. 6b). For individual categories, the model achieved 99% accuracy and 96% recall for n-NP samples, 98% accuracy and 99% recall for NP samples, and 97% accuracy and 99% recall for HC. Crucially, it also accurately predicted subjective pain intensity, with strong correlations between predicted and clinical VAS/NRS scores (R² = 0.961 and 0.926, respectively; Fig. 6c, d). The model-generated connectivity maps (Fig. 6e) revealed distinct network patterns, particularly in parietal-prefrontal and temporo-occipital pathways, providing interpretable neural signatures for each pain state and underscoring NP as a disorder of distributed brain network dynamics. Discussion This study systematically investigates the neural oscillation and functional connectivity patterns that characterize neuropathic pain (NP) using high-density electroencephalography (HD-EEG) combined with graph generative network (GGN) analysis. Our findings robustly demonstrate that NP is not solely a disorder of peripheral nerves but a maladaptive state of the central nervous system, characterized by large-scale dysregulation of neural oscillations and functional networks. These results provide a pathophysiological framework for NP and deliver a suite of objective biomarkers with immediate potential to redefine its diagnosis and assessment. 4.1 Abnormal Low-Frequency Oscillations and Slowed Peak Alpha Frequency Reflect Thalamocortical Dysrhythmia Our investigation demonstrates significant power enhancement in theta (4–8 Hz) and low alpha (8–10 Hz) frequency bands in NP patients during resting state, with predominant distribution in posterior brain regions (occipital and parietal lobes) and prefrontal areas. This observation aligns with the thalamocortical dysrhythmia theory [ 11 ] , which posits that thalamic rhythm disturbances lead to enhanced low-frequency cortical oscillations [ 7 , 12 ] . Particularly noteworthy is the significant slowing of peak alpha frequency (PAF) in NP patients, consistent with previous research suggesting PAF as a potential biomarker for individual pain sensitivity [ 10 ] . Recent studies have further validated the diagnostic value of PAF slowing in various neuropathic pain conditions [ 13 ] . The specific enhancement of prefrontal theta activity carries substantial clinical implications [ 14 ] . The prefrontal cortex serves as a crucial hub for cognitive control, emotional regulation, and descending pain modulation [ 15 ] .The elevated prefrontal theta activity in NP patients may reflect additional cognitive and emotional efforts required to cope with persistent pain signals from abnormal neural discharges [ 7 ] [ 16 ] . This \"neuro-metabolic cost\" potentially encompasses continuous pain monitoring, suppression of pain-related interference, and regulation of negative emotions [ 17 ] . In contrast, non-neuropathic pain patients demonstrate different response patterns in brain cognitive-emotional centers,providing neurophysiological evidence for distinguishing pathophysiological pain types.Recent evidence suggests that these oscillatory patterns may serve as predictive biomarkers for treatment response [ 18 ] . 4.2 Network-Level Functional Reorganization and Information Processing Abnormalities Functional connectivity analysis reveals enhanced whole-brain functional connectivity in theta and low alpha bands among NP patients, particularly within the core \"pain matrix\" network involving prefrontal-anterior cingulate-insula connections. It is worth noting that the mega analysis by Bott et al. [ 19 ] also revealed the most robust association between large-scale brain network connections in the theta band (especially those involving the limbic system) and pain intensity. The study by Li et al [ 20 ] . demonstrated that persistent pain exacerbates individuals' difficulty in disengaging attention from pain-related words and was accompanied by heightened sensitivity of early delta/theta oscillations to \"expectancy violations.\" This aligns closely with our observations of hyperactive theta-band network activity and enhanced theta-gamma coupling.This widespread connectivity enhancement suggests that the NP brain exists in a highly synchronized yet potentially inefficient \"hyperconnected\" state, possibly disrupting the normal balance between network segregation and integration. Of particular significance is our discovery of enhanced cross-frequency coupling (PAC) between occipital theta and prefrontal gamma activity in NP patients.Theta-gamma coupling represents a fundamental mechanism for neural information encoding and transmission [ 21 , 22 ] , where low-frequency theta oscillations modulate the timing of high-frequency gamma activity, thereby supporting working memory, attention, and long-range communication [ 23 , 24 ] .In NP, this enhanced coupling may indicate abnormally increased efficiency in information transmission and binding from sensory integration regions to higher-order cognitive regions in the prefrontal cortex, potentially forming the neural basis for pain memory consolidation, attentional capture, and cognitive bias. Contemporary research has begun to explore the therapeutic potential of modulating these cross-frequency interactions [ 25 , 26 ] . 4.3 A New Paradigm in Pain Assessment: From Subjective Reporting to Objective Brain Network Decoding The Graph-Generative Network (GGN) introduced in this study represents a paradigm shift in pain neuroimaging, moving beyond correlation-based biomarker discovery to establishing a causal-computational framework for objective pain phenotyping. Our model achieves transformative performance in both pain classification (98% accuracy) and intensity prediction, fundamentally redefining what is possible in pain assessment. Critically, GGN transcends the limitations of traditional spectral-feature approaches by directly learning the dynamic spatiotemporal architecture of pain—capturing how distributed brain networks collectively encode the pain experience in real time [ 27 – 29 ] . What distinguishes GGN is its capacity to generate dynamic, interpretable connectivity maps that reveal the precise neural circuits governing different pain states. These maps identify distinct network signatures—particularly in parietal-prefrontal and temporo-occipital pathways—that serve as computational biomarkers for pain subtypes. This represents the first demonstration that subjective pain experiences can be objectively decoded from spontaneous brain activity alone, without provocative stimulation or patient self-report. This work positions GGN as more than an analytical tool; it establishes a new paradigm where brain network dynamics become the primary language for understanding, measuring, and treating pain—finally liberating the field from its centuries-long reliance on subjective reporting and moving us toward a future where pain is understood through the intrinsic organization of brain networks. 4.4 Research Limitations and Clinical Translation Prospects While this study demonstrates significant findings, several limitations warrant consideration. First, the cross-sectional design cannot establish causality, as these neural features could represent either causes, consequences, or both aspects of pain. Second, although non-NP pain control groups were included, potential confounding effects from neuropsychiatric comorbidities (such as anxiety and depression) require further investigation. Additionally, while the sample size is substantial for HD-EEG research, multicenter validation remains necessary to enhance generalizability. Recent methodological reviews have emphasized the importance of addressing these limitations in neuropathic pain research. From a clinical translation perspective, despite current technical complexities in HD-EEG and GGN analysis, rapid advancements in neural technologies are lowering application barriers [ 7 , 30 , 31 ] . Portable EEG devices and cloud computing platforms are making EEG-based pain assessment tools increasingly feasible for clinical practice within the next 5–10 years [ 32 , 33 ] , particularly for assessment and treatment guidance in refractory pain patients [ 34 ] . [ 35 ] Recent technological developments have significantly improved the practicality of EEG-based pain assessment in clinical settings [ 7 , 36 ] . Conclusion This study reveals central mechanism characteristics of NP from both neural oscillation and network connectivity perspectives, establishing a precise identification model based on GGN. These findings not only provide new neural markers for objective assessment of NP but also advance our conceptualization of NP as a brain network disorder. Furthermore, the multidimensional analytical framework and computational methods proposed in this study provide a theoretical foundation for developing targeted neuromodulation therapies, potentially advancing pain medicine toward precision medicine approaches. References Davis KD,Aghaeepour N,Ahn AH,et al.Discovery and validation of biomarkers to aid the development of safe and effective pain therapeutics: challenges and opportunities. Nature Reviews Neurology, 2020. 16 (7): p. 381-400 Kovačević I,Pavić J,Filipović B,et al.Integrated approach to chronic pain—the role of psychosocial factors and multidisciplinary treatment: a narrative review. International journal of environmental research and public health, 2024. 21 (9): p. 1135 Schultz DM,Bakke CH,Ruble HL,et al.Intrathecal drug delivery for intractable pain: identified patient satisfaction survey study comparing intrathecal dose with satisfaction, pain relief, and side effects. Neuromodulation: Technology at the Neural Interface, 2025 Leroux A,Rzasa-Lynn R,Crainiceanu C,et al.Wearable devices: current status and opportunities in pain assessment and management. Digital Biomarkers, 2021. 5 (1): p. 89-102 Ji Y,Ma BJ,Guo XQ,et al.[Discussion on the composition and implementation of diagnosis and treatment strategies for whole field pain management strategy]. Zhonghua Yi Xue Za Zhi, 2023. 103 (39): p. 3083-3087.doi:10.3760/cma.j.cn112137-20230704-01135 Patel M,Hasoon J,Diez Tafur R,et al.The Impact of Chronic Pain on Cognitive Function. Brain Sciences, 2025. 15 (6): p. 559 Ahmad B and Barkana BD.Pain and the Brain: A Systematic Review of Methods, EEG Biomarkers, Limitations, and Future Directions. Neurology International, 2025. 17 (4): p. 46 Mathew J,Perez TM,Adhia DB,et al.Is there a difference in EEG characteristics in acute, chronic, and experimentally induced musculoskeletal pain states? A systematic review. Clinical EEG and neuroscience, 2024. 55 (1): p. 101-120 May ES,Tiemann L,Ávila CG,et al.Assessing the predictive value of peak alpha frequency for the sensitivity to pain. Pain, 2022: p. 10.1097 Chowdhury NS,Bi C,Furman AJ,et al.Predicting individual pain sensitivity using a novel cortical biomarker signature. JAMA neurology, 2025. 82 (3): p. 237-246 Kleeva D,Soghoyan G,Biktimirov A,et al.Modulations in high-density EEG during the suppression of phantom-limb pain with neurostimulation in upper limb amputees. Cerebral Cortex, 2024. 34 (2): p. bhad504 Kannan S,Patel K,Di Basilio D,et al.Shared neural signatures in Functional Neurological Disorder and Chronic Pain: a multimodal narrative review. BMJ Neurology Open, 2025. 7 (2): p. e001032 Cavaleri R,McLain NJ,Heindel M,et al.Peak alpha frequency is related to the degree of widespread pain, but not pain intensity or duration, among people with urologic chronic pelvic pain syndrome. Pain Reports, 2025. 10 (2): p. e1251 Li P,Yokoyama M,Okamoto D,et al.Resting-state EEG features modulated by depressive state in healthy individuals: insights from theta PSD, theta-beta ratio, frontal-parietal PLV, and sLORETA. Frontiers in Human Neuroscience, 2024. 18 : p. 1384330 Li M,She K,Zhu P,et al.Chronic pain and comorbid emotional disorders: neural circuitry and neuroimmunity pathways. International Journal of Molecular Sciences, 2025. 26 (2): p. 436 Chmiel J and Kurpas D.Neural Correlates of Borderline Personality Disorder (BPD) Based on Electroencephalogram (EEG)—A Mechanistic Review. International Journal of Molecular Sciences, 2025. 26 (17): p. 8230 Han S,Wang J,Zhang W,et al.Chronic pain–related cognitive deficits: preclinical insights into molecular, cellular, and circuit mechanisms. Molecular Neurobiology, 2024. 61 (10): p. 8123-8143 Manwatkar S,Puri M,Kumar B,et al.Tracing The Evolution Of Neuropathic Pain Markers: A Journey From Past Discoveries To Emerging Innovations And Future Directions. REDVET-Revista electrónica de Veterinaria. 25 (1): p. 2024 Bott FS,Zebhauser PT,Hohn VD,et al.Exploring electroencephalographic chronic pain biomarkers: a mega-analysis. EBioMedicine, 2025. 120 : p. 105955.doi:10.1016/j.ebiom.2025.105955 Li J,Lyu X,Li X,et al.Pain in focus: How persistent pain disrupts the attentional bias towards pain-related information. Neuroimage, 2025. 321 : p. 121539.doi:10.1016/j.neuroimage.2025.121539 Ursino M and Pirazzini G.Theta–gamma coupling as a ubiquitous brain mechanism: Implications for memory, attention, dreaming, imagination, and consciousness. Current Opinion in Behavioral Sciences, 2024. 59 : p. 101433 Pirazzini G and Ursino M.Modeling the contribution of theta-gamma coupling to sequential memory, imagination, and dreaming. Frontiers in Neural Circuits, 2024. 18 : p. 1326609 Ai H,Zhang S,Si C,et al.Impaired theta and low-gamma directed information flow in the hippocampal-prefrontal circuit underlies working memory deficits in APP/PS1 mice. Behavioral and Brain Functions, 2025. 21 (1): p. 21 Palacino F,Manganotti P and Benussi A.Targeting neural oscillations for cognitive enhancement in Alzheimer’s disease. Medicina, 2025. 61 (3): p. 547 Guo Z,Lin J-P,Simeone O,et al.Cross-frequency cortex–muscle interactions are abnormal in young people with dystonia. Brain Communications, 2024. 6 (2): p. fcae061 Mark JI,Riddle J,Gangwani R,et al.Cross-frequency coupling as a biomarker for early stroke recovery. Neurorehabilitation and neural repair, 2024. 38 (7): p. 506-517 Mohammadi H and Karwowski W.Graph neural networks in brain connectivity studies: Methods, challenges, and future directions. Brain Sciences, 2024. 15 (1): p. 17 Yang H,Huang R,Ye S,et al.A Review of Graph Neural Networks for Brain Diseases Analysis. Available at SSRN 5400612, Zhang W and Hong Q.Modeling brain functional networks using graph neural networks: A review and clinical application. ICCK Transactions on Intelligent Systematics, 2024. 1 (2): p. 58-68 Parsa M,Rad HY,Vaezi H,et al.EEG-based classification of individuals with neuropsychiatric disorders using deep neural networks: A systematic review of current status and future directions. Comput Methods Programs Biomed, 2023. 240 : p. 107683.doi:10.1016/j.cmpb.2023.107683 Pu L,Lion KM,Todorovic M,et al.Portable EEG monitoring for older adults with dementia and chronic pain-a feasibility study. Geriatric Nursing, 2021. 42 (1): p. 124-128 Rice DA,Ozolins C,Biswas R,et al.Home-based EEG neurofeedback for the treatment of chronic pain: a randomized controlled clinical trial. The Journal of Pain, 2024. 25 (11): p. 104651 Teel EF,Ocay DD,Blain-Moraes S,et al.Accurate classification of pain experiences using wearable electroencephalography in adolescents with and without chronic musculoskeletal pain. Frontiers in Pain Research, 2022. 3 : p. 991793 Lovelace JA,Miller J,Jacobs S,et al.Development of Machine Learning Algorithms Using EEG Data to Detect the Presence of Chronic Pain. medRxiv, 2024: p. 2024.2009. 2018.24313903 Chang JL,Nguyen P,Ruan QZ,et al.The Potential of Wearable, Modular Devices in Monitoring Functional Clinical Metrics in Patients Suffering from Chronic Pain. Current Pain and Headache Reports, 2025. 29 (1): p. 46 Alshehri H,Al-Nafjan A and Aldayel M.Decoding Pain: A Comprehensive Review of Computational Intelligence Methods in Electroencephalography-Based Brain–Computer Interfaces. Diagnostics, 2025. 15 (3): p. 300 Additional Declarations There is NO Competing Interest. Supplementary Files SIHighDensityElectroencephalographicAnalysisofNeuropathicPainAProspectiveMulticenterClinicalStudy.docx SI-High-Density Electroencephalographic Analysis of Neuropathic Pain - A Prospective, Multicenter Clinical Study Cite Share Download PDF Status: Under Review Version 1 posted You are reading this latest preprint version Research Square lets you share your work early, gain feedback from the community, and start making changes to your manuscript prior to peer review in a journal. As a division of Research Square Company, we’re committed to making research communication faster, fairer, and more useful. We do this by developing innovative software and high quality services for the global research community. Our growing team is made up of researchers and industry professionals working together to solve the most critical problems facing scientific publishing. Also discoverable on Platform About Our Team In Review Editorial Policies Advisory Board Help Center Resources Author Services Accessibility API Access RSS feed Manage Cookie Preferences © Research Square 2026 | ISSN 2693-5015 (online) Privacy Policy Terms of Service Do Not Sell My Personal Information {\"props\":{\"pageProps\":{\"initialData\":{\"identity\":\"rs-8008823\",\"acceptedTermsAndConditions\":true,\"allowDirectSubmit\":false,\"archivedVersions\":[],\"articleType\":\"Article\",\"associatedPublications\":[],\"authors\":[{\"id\":539918072,\"identity\":\"d70adb23-e23c-46d1-a7b2-4890a5cafcf5\",\"order_by\":0,\"name\":\"Ke Ma\",\"email\":\"data:image/png;base64,iVBORw0KGgoAAAANSUhEUgAAAZAAAAAyAQMAAABI0h/eAAAABlBMVEX///8AAABVwtN+AAAACXBIWXMAAA7EAAAOxAGVKw4bAAAA2UlEQVRIie3PsQrCMBCA4SuBuARcs2ifQKgURKgPkyIki3V2cIiLTuLazVfoI0QCThXXriJ07iqIGnFwi3ETzL8cgfvgAuDz/WYIgtuoC6BeDzcCmMcA+juiU/nadiC91eEcNQSJ7eK4pzBLUtk6KCsZlCJOc4qzQmlMoRSpJFNmJ4ojTSKSFWBIsDQXUhLZybFG+sqoCOWT3FxIxdEYVMTgeVggnUiN+mvJ+uYv8ZDtRbwkk0+HcUQv8h6G+e5UNfOks2mVdvKOKgBmJnbcN7Wl+67P5/P9Vw+1WER61ihFkwAAAABJRU5ErkJggg==\",\"orcid\":\"\",\"institution\":\"Ninth People's Hospital Affiliated to Shanghai Jiao Tong University School of Medicine, Shanghai\",\"correspondingAuthor\":true,\"prefix\":\"\",\"firstName\":\"Ke\",\"middleName\":\"\",\"lastName\":\"Ma\",\"suffix\":\"\"},{\"id\":539918073,\"identity\":\"51c5a96a-314e-4aa5-b1fa-2065b8178bec\",\"order_by\":1,\"name\":\"Yun Ji\",\"email\":\"\",\"orcid\":\"\",\"institution\":\"Department of Pain Medicine, Ninth People's Hospital Affiliated to Shanghai Jiao Tong University School of Medicine\",\"correspondingAuthor\":false,\"prefix\":\"\",\"firstName\":\"Yun\",\"middleName\":\"\",\"lastName\":\"Ji\",\"suffix\":\"\"},{\"id\":539918074,\"identity\":\"7be18739-6b89-4b43-9de9-002f0699c97e\",\"order_by\":2,\"name\":\"Zhiyuan Dang\",\"email\":\"\",\"orcid\":\"\",\"institution\":\"Shanghai Jiao Tong University\",\"correspondingAuthor\":false,\"prefix\":\"\",\"firstName\":\"Zhiyuan\",\"middleName\":\"\",\"lastName\":\"Dang\",\"suffix\":\"\"},{\"id\":539918075,\"identity\":\"e3f40e77-db99-4285-b1f7-e6fb74a66aae\",\"order_by\":3,\"name\":\"Huichun Luo\",\"email\":\"\",\"orcid\":\"\",\"institution\":\"Shanghai Mental Health Center\",\"correspondingAuthor\":false,\"prefix\":\"\",\"firstName\":\"Huichun\",\"middleName\":\"\",\"lastName\":\"Luo\",\"suffix\":\"\"},{\"id\":539918076,\"identity\":\"9f7e16cd-1eed-401c-8d93-d8609a8b94cb\",\"order_by\":4,\"name\":\"Tao Li\",\"email\":\"\",\"orcid\":\"\",\"institution\":\"Department of orthopedics School of Medicine, Xin Hua Hospital Affiliated to Shanghai Jiao Tong University, 1665 Kangjiang Road, Shanghai 200092,\",\"correspondingAuthor\":false,\"prefix\":\"\",\"firstName\":\"Tao\",\"middleName\":\"\",\"lastName\":\"Li\",\"suffix\":\"\"},{\"id\":539918077,\"identity\":\"2140d4b2-aff2-4982-a53a-d6ae04f9d9d7\",\"order_by\":5,\"name\":\"Yongyong Ren\",\"email\":\"\",\"orcid\":\"https://orcid.org/0000-0001-9217-3483\",\"institution\":\"Shanghai Jiao Tong University\",\"correspondingAuthor\":false,\"prefix\":\"\",\"firstName\":\"Yongyong\",\"middleName\":\"\",\"lastName\":\"Ren\",\"suffix\":\"\"},{\"id\":539918078,\"identity\":\"3787e80d-63a1-4a33-8785-bb24a7b9b12f\",\"order_by\":6,\"name\":\"Yufan Ye\",\"email\":\"\",\"orcid\":\"\",\"institution\":\"Department of Bioinformatics and Biostatistics, School of Life Sciences and Biotechnology, Shanghai Jiao Tong University, Shanghai\",\"correspondingAuthor\":false,\"prefix\":\"\",\"firstName\":\"Yufan\",\"middleName\":\"\",\"lastName\":\"Ye\",\"suffix\":\"\"},{\"id\":539918079,\"identity\":\"cdca0d01-8576-4712-afd2-c23b8a329a66\",\"order_by\":7,\"name\":\"Zhenyu Cai\",\"email\":\"\",\"orcid\":\"\",\"institution\":\"The School of Clinical Medicine, Fujian Medical University\",\"correspondingAuthor\":false,\"prefix\":\"\",\"firstName\":\"Zhenyu\",\"middleName\":\"\",\"lastName\":\"Cai\",\"suffix\":\"\"},{\"id\":539918080,\"identity\":\"6aa7c156-1610-4dc1-ae9a-d88909c272d4\",\"order_by\":8,\"name\":\"Tifei Yuan\",\"email\":\"\",\"orcid\":\"https://orcid.org/0000-0003-0510-715X\",\"institution\":\"Shanghai Jiaotong University\",\"correspondingAuthor\":false,\"prefix\":\"\",\"firstName\":\"Tifei\",\"middleName\":\"\",\"lastName\":\"Yuan\",\"suffix\":\"\"},{\"id\":539918081,\"identity\":\"abd7488c-e372-4fc1-bc61-907d77c882cd\",\"order_by\":9,\"name\":\"Yue Zhang\",\"email\":\"\",\"orcid\":\"\",\"institution\":\"Shanghai Jiao Tong University\",\"correspondingAuthor\":false,\"prefix\":\"\",\"firstName\":\"Yue\",\"middleName\":\"\",\"lastName\":\"Zhang\",\"suffix\":\"\"},{\"id\":539918082,\"identity\":\"456ab16c-9842-4e1f-baf4-cbbd6af8fb50\",\"order_by\":10,\"name\":\"Pingying Wang\",\"email\":\"\",\"orcid\":\"\",\"institution\":\"Department of Pain, The First Affiliated Hospital of China Medical University\",\"correspondingAuthor\":false,\"prefix\":\"\",\"firstName\":\"Pingying\",\"middleName\":\"\",\"lastName\":\"Wang\",\"suffix\":\"\"},{\"id\":539918083,\"identity\":\"939d041d-2557-408c-ae3b-f909de4697c4\",\"order_by\":11,\"name\":\"Zhenxing Li\",\"email\":\"\",\"orcid\":\"\",\"institution\":\"Department of Pain, The Third Xiangya Hospital and Institute of Pain Medicine, Central South University\",\"correspondingAuthor\":false,\"prefix\":\"\",\"firstName\":\"Zhenxing\",\"middleName\":\"\",\"lastName\":\"Li\",\"suffix\":\"\"}],\"badges\":[],\"createdAt\":\"2025-11-02 04:40:10\",\"currentVersionCode\":1,\"declarations\":\"\",\"doi\":\"10.21203/rs.3.rs-8008823/v1\",\"doiUrl\":\"https://doi.org/10.21203/rs.3.rs-8008823/v1\",\"draftVersion\":[],\"editorialEvents\":[],\"editorialNote\":\"\",\"failedWorkflow\":false,\"files\":[{\"id\":97342971,\"identity\":\"e4beaca0-13f9-4c94-87f9-f090294bfd21\",\"added_by\":\"auto\",\"created_at\":\"2025-12-03 11:36:09\",\"extension\":\"png\",\"order_by\":1,\"title\":\"Figure 1\",\"display\":\"\",\"copyAsset\":false,\"role\":\"figure\",\"size\":2000281,\"visible\":true,\"origin\":\"\",\"legend\":\"\\u003cp\\u003eThe research and main analysis steps of this design are summarized. This study employed three sets of electroencephalogram (EEG) datasets. High-density EEG was recorded in both open-eye (EO) and closed-eye (EC) modes. Source localization was used to estimate the activities of cortical and subcortical regions, and the power functional connectivity characteristics among multiple regions were calculated to reveal the neuropathological mechanisms of chronic pain and train predictive models for different subtypes of chronic pain. Based on the dataset, a new type of graph generation network (GGN) was developed to simulate the dynamic probability of functional connections between electrode nodes by capturing the inherent nonlinear spatial and temporal features in high-density electroencephalogram signals, accurately predicting the intensity of pain.\\u003c/p\\u003e\",\"description\":\"\",\"filename\":\"image1.png\",\"url\":\"https://assets-eu.researchsquare.com/files/rs-8008823/v1/7fef846f8b2b1c96e3d5252f.png\"},{\"id\":97342972,\"identity\":\"18a7b455-e3d4-448f-97e8-23c1e54ed4ef\",\"added_by\":\"auto\",\"created_at\":\"2025-12-03 11:36:09\",\"extension\":\"png\",\"order_by\":2,\"title\":\"Figure 2\",\"display\":\"\",\"copyAsset\":false,\"role\":\"figure\",\"size\":7539054,\"visible\":true,\"origin\":\"\",\"legend\":\"\\u003cp\\u003e(a) Averaged power spectral density (PSD) curves across all electrodes for the three groups. The 95% confidence interval and peak alpha frequency (PAF) are indicated for each curve, with gray shading indicating frequency bands with strong visual differences. (b) Boxplots of mean of band-limited power across six frequency bands for each group, averaged across all electrodes. Statistical comparisons were conducted using the adjusted t-test (FDR).\\u003c/p\\u003e\",\"description\":\"\",\"filename\":\"image2.png\",\"url\":\"https://assets-eu.researchsquare.com/files/rs-8008823/v1/d98592dbe8b78d559bed9dde.png\"},{\"id\":97342967,\"identity\":\"35f7dd42-f64b-40f0-aaff-5c739c35976f\",\"added_by\":\"auto\",\"created_at\":\"2025-12-03 11:36:08\",\"extension\":\"png\",\"order_by\":3,\"title\":\"Figure 3\",\"display\":\"\",\"copyAsset\":false,\"role\":\"figure\",\"size\":7532221,\"visible\":true,\"origin\":\"\",\"legend\":\"\\u003cp\\u003e(a) Topographic maps showing PSD. Warmer colors represent higher power. (b) Topographic maps showing pairwise PSD differences between groups for six frequency band at all electrodes. Warmer colors represent higher Tukey’s Q-statistic, indicating significantly greater activity in the preceding group. (c) Tukey's Q-statistical distribution map of electrodes from different brain regions compared pairwise, in the theta (4-8 Hz) band and low-alpha (8-10 Hz) band. The dashed line represents the threshold of significance level α=0.05.\\u003c/p\\u003e\",\"description\":\"\",\"filename\":\"image3.png\",\"url\":\"https://assets-eu.researchsquare.com/files/rs-8008823/v1/d1f13da0eab968030f390ce1.png\"},{\"id\":97342970,\"identity\":\"2ecdb8c4-e799-4f2f-ab3a-a199951a6a7f\",\"added_by\":\"auto\",\"created_at\":\"2025-12-03 11:36:08\",\"extension\":\"png\",\"order_by\":4,\"title\":\"Figure 4\",\"display\":\"\",\"copyAsset\":false,\"role\":\"figure\",\"size\":1876187,\"visible\":true,\"origin\":\"\",\"legend\":\"\\u003cp\\u003e(a) Boxplots of band average power for the top three electrodes in terms of SHAP value. (b) Confusion matrix for the three-class classification task using a LASSO model. Trained with spectral features, which includes the average power characteristics of six frequency bands for all electrodes. (c) Scatter plot of SHAP value feature importance for each non-zero weighted feature point in each class. The larger the positive SHAP value, the greater the positive effect of the feature in predicting the class.The top 1% features and corresponding classes of SHAP value's absolute value ranking are plotted in the upper left corner.\\u003c/p\\u003e\",\"description\":\"\",\"filename\":\"image4.png\",\"url\":\"https://assets-eu.researchsquare.com/files/rs-8008823/v1/4be8264ae9aeaca9a5a9dce0.png\"},{\"id\":97342969,\"identity\":\"75028403-f1b2-4731-a014-02dbcc80c5d4\",\"added_by\":\"auto\",\"created_at\":\"2025-12-03 11:36:08\",\"extension\":\"png\",\"order_by\":5,\"title\":\"Figure 5\",\"display\":\"\",\"copyAsset\":false,\"role\":\"figure\",\"size\":5868641,\"visible\":true,\"origin\":\"\",\"legend\":\"\\u003cp\\u003e(a) Histograms of the average connectivity values in different frequency bands of groups. (b) Brain maps showing the adjusted Tukey's Q-statistic of wPLI between two groups of Neuropathic pain patients and healthy controls in theta band (4-8 Hz), with Bonferroni-corrected p \\u0026lt; 0.01. (c) The adjusted Tukey's Q-statistic of wPLI between Not-neuropathic pain patients and healthy controls in theta band (4-8 Hz). (d) Phase-amplitude coupling (PAC) of the theta band of the occipital lobe to the gamma band (30-75 Hz) of the frontal lobe, line colors indicate the modulation index (MI).\\u003c/p\\u003e\",\"description\":\"\",\"filename\":\"image5.png\",\"url\":\"https://assets-eu.researchsquare.com/files/rs-8008823/v1/4ac8aa5d4667bc1e18628c79.png\"},{\"id\":97342965,\"identity\":\"2e521156-41ec-4be8-8983-c5ce276a7810\",\"added_by\":\"auto\",\"created_at\":\"2025-12-03 11:36:08\",\"extension\":\"png\",\"order_by\":6,\"title\":\"Figure 6\",\"display\":\"\",\"copyAsset\":false,\"role\":\"figure\",\"size\":217206,\"visible\":true,\"origin\":\"\",\"legend\":\"\\u003cp\\u003e(a) The architecture diagram of the Graph-generative Network (GGN) model. (b) The confusion matrix of the classification result. (c) Regression prediction of VAS score for pain patients. (d) Regression prediction of NRS score for pain patients. (e) The brain connectivity estimated by the connectivity graph generator of the GGN model, which reflects the importance of the connections between different regions for the model to classify the test set into specific categories.\\u003c/p\\u003e\",\"description\":\"\",\"filename\":\"image6.png\",\"url\":\"https://assets-eu.researchsquare.com/files/rs-8008823/v1/78aab28b6fea0f5def0c4449.png\"},{\"id\":97664591,\"identity\":\"e2ebf353-297b-453b-bf96-616ed6b9f785\",\"added_by\":\"auto\",\"created_at\":\"2025-12-08 09:11:04\",\"extension\":\"pdf\",\"order_by\":0,\"title\":\"\",\"display\":\"\",\"copyAsset\":false,\"role\":\"manuscript-pdf\",\"size\":22022976,\"visible\":true,\"origin\":\"\",\"legend\":\"\",\"description\":\"\",\"filename\":\"manuscript.pdf\",\"url\":\"https://assets-eu.researchsquare.com/files/rs-8008823/v1/55e4e4ba-84b0-4549-97d9-1718a7514741.pdf\"},{\"id\":97369469,\"identity\":\"898cadc3-7eca-4818-a758-541821941fc1\",\"added_by\":\"auto\",\"created_at\":\"2025-12-03 16:25:00\",\"extension\":\"docx\",\"order_by\":1,\"title\":\"\",\"display\":\"\",\"copyAsset\":false,\"role\":\"supplement\",\"size\":6083562,\"visible\":true,\"origin\":\"\",\"legend\":\"SI-High-Density Electroencephalographic Analysis of Neuropathic Pain - A Prospective, Multicenter Clinical Study\",\"description\":\"\",\"filename\":\"SIHighDensityElectroencephalographicAnalysisofNeuropathicPainAProspectiveMulticenterClinicalStudy.docx\",\"url\":\"https://assets-eu.researchsquare.com/files/rs-8008823/v1/2a5555cf02d923303a105ba1.docx\"}],\"financialInterests\":\"There is \\u003cb\\u003eNO\\u003c/b\\u003e Competing Interest.\",\"formattedTitle\":\"High-Density Electroencephalographic Analysis of Neuropathic Pain - A Prospective, Multicenter Clinical Study\",\"fulltext\":[{\"header\":\"Introduction\",\"content\":\"\\u003cp\\u003eA critical challenge facing pain medicine is the severe lack of objective tools for assessing pain severity and monitoring treatment efficacy\\u003csup\\u003e[\\u003cspan citationid=\\\"CR1\\\" class=\\\"CitationRef\\\"\\u003e1\\u003c/span\\u003e]\\u003c/sup\\u003e. Despite the advancements in drug therapies, minimally invasive interventions, and multidisciplinary collaborations aimed at tackling chronic pain, research and surveys have shown a troubling trend of high treatment dissatisfaction rates among chronic pain patients\\u003csup\\u003e[\\u003cspan citationid=\\\"CR2\\\" class=\\\"CitationRef\\\"\\u003e2\\u003c/span\\u003e]\\u003c/sup\\u003e. Based on the latest research and surveys, the dissatisfaction rate of chronic pain patients with analgesic effects is as high as 64%, and the rate of complete relief after treatment for chronic pain patients is less than 20%\\u003csup\\u003e[\\u003cspan citationid=\\\"CR3\\\" class=\\\"CitationRef\\\"\\u003e3\\u003c/span\\u003e]\\u003c/sup\\u003e. This persistent therapeutic gap is largely attributable to the reliance on subjective patient-reported outcomes like the Visual Analog Scale (VAS) and Numeric Rating Scale (NRS), which fail to provide consistent, quantitative, and biomarker-independent measures of pain experience\\u003csup\\u003e[\\u003cspan citationid=\\\"CR4\\\" class=\\\"CitationRef\\\"\\u003e4\\u003c/span\\u003e]\\u003c/sup\\u003e. Furthermore, existing clinical approaches often narrowly target the sensory dimension of pain, overlooking its complex emotional, cognitive, and behavioral aspects\\u003csup\\u003e[\\u003cspan citationid=\\\"CR5\\\" class=\\\"CitationRef\\\"\\u003e5\\u003c/span\\u003e]\\u003c/sup\\u003e.The brain serves as the foundation of human cognition, emotion, and ultimately, pain perception, which involves intricate neural networks. A growing body of evidence indicates that chronic pain is linked to extensive and measurable changes in brain structure and function\\u003csup\\u003e[\\u003cspan citationid=\\\"CR6\\\" class=\\\"CitationRef\\\"\\u003e6\\u003c/span\\u003e]\\u003c/sup\\u003e. Understanding these neurophysiological signatures is crucial not only for deciphering the pathophysiology of pain but also for developing the long-sought objective biomarkers essential for precise diagnosis and effective treatment.\\u003c/p\\u003e\\u003cp\\u003eElectroencephalography (EEG), which non-invasively records the brain's electrophysiological activity, has long served as a valuable tool in pain research \\u003csup\\u003e[\\u003cspan citationid=\\\"CR7\\\" class=\\\"CitationRef\\\"\\u003e7\\u003c/span\\u003e]\\u003c/sup\\u003e. However, conventional EEG methodologies have predominantly relied on stimulus-evoked paradigms\\u0026mdash;typically using acute, experimentally induced pain\\u0026mdash;which suffer from critical limitations in the context of chronic pain. These approaches are not only clinically impractical due to requirements for standardized stimulation and patient cooperation, but also fail to capture the persistent, resting-state brain activity that characterizes chronic neuropathic conditions \\u003csup\\u003e[\\u003cspan citationid=\\\"CR7\\\" class=\\\"CitationRef\\\"\\u003e7\\u003c/span\\u003e, \\u003cspan citationid=\\\"CR8\\\" class=\\\"CitationRef\\\"\\u003e8\\u003c/span\\u003e]\\u003c/sup\\u003e. Furthermore, traditional low-density EEG systems, often limited to 32 or fewer electrodes, provide insufficient spatial sampling to resolve the distributed network dynamics underlying pain processing. The brain's complex neural architecture, involving intricate signal transmissions across widely distributed regions such as the prefrontal cortex, anterior cingulate, insula, and somatosensory areas, necessitates high-density spatial sampling to accurately map functional interactions within the pain neuromatrix. This fundamental limitation in spatial resolution has likely contributed to inconsistencies across previous studies and hindered the identification of reliable, clinically applicable neural signatures for chronic neuropathic pain.\\u003c/p\\u003e\\u003cp\\u003eThis study addresses these critical limitations. Leveraging high-density EEG (HD-EEG), which employs a significantly greater number of densely positioned electrodes, enables a far more comprehensive recording of the brain's electrical signals and network interactions. Crucially, our approach focuses on resting-state brain activity, eliminating the need for artificial pain provocation and enhancing clinical feasibility. Most significantly, we integrate these rich HD-EEG datasets with advanced computational approaches, particularly Graph Neural Network (GNN)-based models like our novel Graph-Generative Network (GGN). These powerful algorithms are designed to learn the complex spatial and temporal patterns inherent in the EEG data. This enables not only the highly accurate classification of pain types (e.g., neuropathic vs. non-neuropathic) directly from resting-state signals but also the precise prediction of subjective pain intensity levels, offering a transformative pathway towards objective, quantitative pain assessment. Conducted within the China Chronic Pain Cohort (CPCC), the largest prospective, multicenter clinical cohort of its kind in China to date, this research provides robust evidence derived from a uniquely representative sample. By observing these detailed resting-state activity patterns across the brain in this extensive cohort, researchers can more accurately identify and comprehend the neural pathways associated with chronic pain perception and its modulation, paving the way for improved management strategies.\\u003c/p\\u003e\"},{\"header\":\"Results\",\"content\":\"\\u003cp\\u003e\\u003cstrong\\u003e3.1 Experimental design and analytical workflow of the HD-EEG study.\\u003c/strong\\u003e\\u003c/p\\u003e\\n\\u003cp\\u003eOur high-density electroencephalography (HD-EEG) analysis of the China Chronic Pain Cohort identified robust and replicable neurophysiological signatures that not only distinguish chronic pain patients from healthy controls but also permit precise subtyping of pain conditions. As illustrated in F\\u003cstrong\\u003eig. 1\\u003c/strong\\u003e, we analyzed resting-state neural activity from 354 participants across both eyes-open and eyes-closed conditions, thereby capturing state-dependent cortical dynamics relevant to sensory processing and internal cognitive states. The analytical pipeline was structured as a hierarchically integrated computational workflow: we first conducted source reconstruction using the Brainstorm toolbox to localize group-level differences in cortical activity and examine their relationship with neuropathic pain; we then quantified spectral power and functional connectivity\\u0026mdash;including weighted phase lag index (wPLI) and phase-amplitude coupling\\u0026mdash;to characterize oscillatory profiles and network-level interactions; finally, we introduced a novel Graph-Generative Network (GGN) that integrates these spatiotemporal features to extract pain-specific biomarkers and objectively predict clinical pain intensity directly from neural activity, achieving state-of-the-art performance in both diagnostic classification and pain scoring.\\u0026nbsp;\\u003c/p\\u003e\\n\\u003cp\\u003e\\u003cstrong\\u003e3.2 The average pow\\u003c/strong\\u003e\\u003cstrong\\u003eer\\u0026nbsp;\\u003c/strong\\u003e\\u003cstrong\\u003espectra\\u003c/strong\\u003e\\u003cstrong\\u003e\\u0026nbsp;across all electrodes\\u003c/strong\\u003e\\u003c/p\\u003e\\n\\u003cp\\u003eSpectral analysis revealed fundamental oscillatory abnormalities in NP patients, forming the basis for subsequent spatial and network-level investigations. As shown in Fig. 2a, the averaged power spectral density (PSD) curves exhibited notable group differences, with NP patients demonstrating significant power enhancements in the theta (4\\u0026ndash;8 Hz) and low-alpha (8\\u0026ndash;10 Hz) bands. Statistical analysis using one-way ANOVA with FDR correction confirmed these observations across six frequency bands (Fig. 2b), revealing significant group differences in theta (F(2,241) = 7.874, p = 0.000487, \\u0026eta;\\u0026sup2; = 0.0613) and low-alpha bands (F(2,241) = 6.302, p = 0.00215, \\u0026eta;\\u0026sup2; = 0.0497). Critically, the peak alpha frequency (PAF) was significantly slowed in NP patients compared to both n-NP patients and healthy controls (HC) (FDR = 0.0000598, Cohen\\u0026apos;s d = -0.604) (Fig. 2c, d), aligning with the thalamocortical dysrhythmia theory.\\u003c/p\\u003e\\n\\u003cp\\u003eIt is worth noting that the relationship of PAF with pain sensitivity and chronic pain has been pointed out in several studies, such as the inability of acute irritant pain to induce changes in PAF in subjects\\u003csup\\u003e[9]\\u003c/sup\\u003e, while in chronic pain, PAF can be used as a biomarker of individual pain sensitivity, and slower PAF can predict higher pain levels\\u003csup\\u003e[10]\\u003c/sup\\u003e. In our study, both types of chronic pain resulted in the slowest PAF, and neuropathic pain, which was more painful, had the slowest PAF, which was associated with increased pain, this shows the potential of PAF indicators as biomarkers in chronic pain.\\u003c/p\\u003e\\n\\u003cp\\u003e\\u003cstrong\\u003e3.3 Spectral related spatial distribution\\u003c/strong\\u003e\\u003c/p\\u003e\\n\\u003cp\\u003eBuilding upon these spectral abnormalities, we mapped their spatial distribution to pinpoint the specific brain regions involved. The topographic maps in Fig. 3a illustrate the group-level spatial distribution of mean power, revealing enhanced theta and low-alpha power in NP patients, with a notable predominance in posterior regions. Fig. 3b displays pairwise PSD differences between groups, quantified by Tukey\\u0026apos;s Q-statistic, highlighting significant enhancements in NP patients compared to HC across most brain regions.\\u0026nbsp;\\u003c/p\\u003e\\n\\u003cp\\u003eThe statistical distribution in Fig. 3c confirms that in the theta band, NP patients exhibited significantly higher power than HC, with the most pronounced enhancements in the occipital lobe and posterior aspects of the temporal lobes (p \\u0026lt; 0.001). Electrode-level analysis identified the most significant enhancements at P7 (FDR = 0.0000166), O1 (FDR = 0.0000214), and O2 (FDR = 0.0000464), all located at the occipital-parietal junction. In the low-alpha band, NP patients showed significantly higher power than HC across the entire occipital lobe, as well as parts of the prefrontal and temporal lobes (p \\u0026lt; 0.05), with maximal differences at Oz (p \\u0026lt; 0.001).\\u003c/p\\u003e\\n\\u003cp\\u003e\\u003cstrong\\u003eHigh-Density Electroencephalographic Analysis of Neuropathic Pain - A Prospective, Multicenter Clinical Study\\u003c/strong\\u003e\\u003c/p\\u003e\\n\\u003cp\\u003e\\u003cstrong\\u003e3.4 Machine-Learning classifier Based on spatial spectral differences\\u003c/strong\\u003e\\u003c/p\\u003e\\n\\u003cp\\u003eLeveraging these spatially distinct spectral features, we developed a LASSO logistic regression model to evaluate their collective power for objective pain classification. Our feature space comprised 762 spectral features (127 electrodes \\u0026times; 6 frequency bands) from 3,125 EEG segments.\\u0026nbsp;\\u003c/p\\u003e\\n\\u003cp\\u003eFig. 4a presents the band average power for the top three electrodes with the highest SHAP values, underscoring the importance of specific electrodes in the theta and low-alpha bands for classification. The LASSO model with optimal regularization (\\u0026lambda; = 0.05) retained 107 features (14.4% of original features), achieving an overall accuracy of 83% in the three-class classification task (Fig. 4b), with particularly high performance in identifying NP samples (accuracy = 0.85, recall = 0.88). Fig. 4c illustrates the SHAP value feature importance, confirming that the theta and low-alpha bands were the most critical for classification (mean SHAP: 0.518 and 0.571, respectively), consistent with our univariate analyses. The model demonstrated AUC values of 0.685 for NP versus other categories, 0.605 for n-NP, and 0.723 for HC.\\u003c/p\\u003e\\n\\u003cp\\u003e\\u003cstrong\\u003e3.5 Brain area coherence and functional connectivity\\u003c/strong\\u003e\\u003c/p\\u003e\\n\\u003cp\\u003eTo investigate whether NP involves large-scale network dysfunction beyond local spectral changes, we analyzed functional connectivity and cross-frequency interactions using the weighted phase lag index (wPLI) and phase-amplitude coupling (PAC). Fig. 5a shows the average wPLI connectivity strength across different frequency bands for the three groups, indicating enhanced theta and low-alpha band connectivity in NP patients. wPLI analysis demonstrated widespread enhancements in functional connectivity for NP patients, most prominently in the theta band (336 connections significantly enhanced vs. HC, Bonferroni-corrected p \\u0026lt; 0.01).\\u0026nbsp;\\u003c/p\\u003e\\n\\u003cp\\u003eFig. 5b and 5c visualize the significant connectivity differences in the theta band between NP patients and HC, and between n-NP patients and HC, respectively, with key strengthened pathways involving prefrontal, cingulate, and insular regions. Notable enhanced connections included those between the left caudal anterior cingulate cortex and right precuneus (p = 0.000011), and between the right isthmus cingulate cortex and left rostral middle frontal gyrus (p = 0.0000098). Phase-amplitude coupling (PAC) analysis further revealed enhanced theta-gamma coupling between occipital and frontal regions in NP (Fig. 5d), suggesting altered cross-frequency communication underlying pain processing.\\u003c/p\\u003e\\n\\u003cp\\u003e\\u003cstrong\\u003e3.6 Graph Generative Network\\u003c/strong\\u003e\\u003cstrong\\u003e\\u0026nbsp;(GGN)\\u003c/strong\\u003e\\u003c/p\\u003e\\n\\u003cp\\u003eFinally, to integrate these local and network-level features into a unified, clinically translatable model, we developed a Graph-Generative Network (GGN) to dynamically model brain connectivity from HD-EEG signals. The GGN architecture is depicted in Fig. 6a, featuring a novel connectivity graph generator that learns dynamic functional connectivity patterns from 127-channel EEG data. The model achieved exceptional performance, differentiating NP, n-NP, and HC with an overall accuracy of 98%, substantially outperforming the LASSO classifier (Fig. 6b).\\u0026nbsp;\\u003c/p\\u003e\\n\\u003cp\\u003eFor individual categories, the model achieved 99% accuracy and 96% recall for n-NP samples, 98% accuracy and 99% recall for NP samples, and 97% accuracy and 99% recall for HC. Crucially, it also accurately predicted subjective pain intensity, with strong correlations between predicted and clinical VAS/NRS scores (R\\u0026sup2; = 0.961 and 0.926, respectively; Fig. 6c, d). The model-generated connectivity maps (Fig. 6e) revealed distinct network patterns, particularly in parietal-prefrontal and temporo-occipital pathways, providing interpretable neural signatures for each pain state and underscoring NP as a disorder of distributed brain network dynamics.\\u003c/p\\u003e\"},{\"header\":\"Discussion\",\"content\":\"\\u003cp\\u003eThis study systematically investigates the neural oscillation and functional connectivity patterns that characterize neuropathic pain (NP) using high-density electroencephalography (HD-EEG) combined with graph generative network (GGN) analysis. Our findings robustly demonstrate that NP is not solely a disorder of peripheral nerves but a maladaptive state of the central nervous system, characterized by large-scale dysregulation of neural oscillations and functional networks. These results provide a pathophysiological framework for NP and deliver a suite of objective biomarkers with immediate potential to redefine its diagnosis and assessment.\\u003c/p\\u003e\\u003cdiv id=\\\"Sec10\\\" class=\\\"Section2\\\"\\u003e\\u003ch2\\u003e4.1 Abnormal Low-Frequency Oscillations and Slowed Peak Alpha Frequency Reflect Thalamocortical Dysrhythmia\\u003c/h2\\u003e\\u003cp\\u003eOur investigation demonstrates significant power enhancement in theta (4\\u0026ndash;8 Hz) and low alpha (8\\u0026ndash;10 Hz) frequency bands in NP patients during resting state, with predominant distribution in posterior brain regions (occipital and parietal lobes) and prefrontal areas. This observation aligns with the thalamocortical dysrhythmia theory \\u003csup\\u003e[\\u003cspan citationid=\\\"CR11\\\" class=\\\"CitationRef\\\"\\u003e11\\u003c/span\\u003e]\\u003c/sup\\u003e, which posits that thalamic rhythm disturbances lead to enhanced low-frequency cortical oscillations \\u003csup\\u003e[\\u003cspan citationid=\\\"CR7\\\" class=\\\"CitationRef\\\"\\u003e7\\u003c/span\\u003e, \\u003cspan citationid=\\\"CR12\\\" class=\\\"CitationRef\\\"\\u003e12\\u003c/span\\u003e]\\u003c/sup\\u003e. Particularly noteworthy is the significant slowing of peak alpha frequency (PAF) in NP patients, consistent with previous research suggesting PAF as a potential biomarker for individual pain sensitivity \\u003csup\\u003e[\\u003cspan citationid=\\\"CR10\\\" class=\\\"CitationRef\\\"\\u003e10\\u003c/span\\u003e]\\u003c/sup\\u003e. Recent studies have further validated the diagnostic value of PAF slowing in various neuropathic pain conditions\\u003csup\\u003e[\\u003cspan citationid=\\\"CR13\\\" class=\\\"CitationRef\\\"\\u003e13\\u003c/span\\u003e]\\u003c/sup\\u003e .\\u003c/p\\u003e\\u003cp\\u003eThe specific enhancement of prefrontal theta activity carries substantial clinical implications\\u003csup\\u003e[\\u003cspan citationid=\\\"CR14\\\" class=\\\"CitationRef\\\"\\u003e14\\u003c/span\\u003e]\\u003c/sup\\u003e. The prefrontal cortex serves as a crucial hub for cognitive control, emotional regulation, and descending pain modulation\\u003csup\\u003e[\\u003cspan citationid=\\\"CR15\\\" class=\\\"CitationRef\\\"\\u003e15\\u003c/span\\u003e]\\u003c/sup\\u003e.The elevated prefrontal theta activity in NP patients may reflect additional cognitive and emotional efforts required to cope with persistent pain signals from abnormal neural discharges\\u003csup\\u003e[\\u003cspan citationid=\\\"CR7\\\" class=\\\"CitationRef\\\"\\u003e7\\u003c/span\\u003e] [\\u003cspan citationid=\\\"CR16\\\" class=\\\"CitationRef\\\"\\u003e16\\u003c/span\\u003e]\\u003c/sup\\u003e. This \\\"neuro-metabolic cost\\\" potentially encompasses continuous pain monitoring, suppression of pain-related interference, and regulation of negative emotions \\u003csup\\u003e[\\u003cspan citationid=\\\"CR17\\\" class=\\\"CitationRef\\\"\\u003e17\\u003c/span\\u003e]\\u003c/sup\\u003e. In contrast, non-neuropathic pain patients demonstrate different response patterns in brain cognitive-emotional centers,providing neurophysiological evidence for distinguishing pathophysiological pain types.Recent evidence suggests that these oscillatory patterns may serve as predictive biomarkers for treatment response \\u003csup\\u003e[\\u003cspan citationid=\\\"CR18\\\" class=\\\"CitationRef\\\"\\u003e18\\u003c/span\\u003e]\\u003c/sup\\u003e.\\u003c/p\\u003e\\u003c/div\\u003e\\u003cdiv id=\\\"Sec11\\\" class=\\\"Section2\\\"\\u003e\\u003ch2\\u003e4.2 Network-Level Functional Reorganization and Information Processing Abnormalities\\u003c/h2\\u003e\\u003cp\\u003eFunctional connectivity analysis reveals enhanced whole-brain functional connectivity in theta and low alpha bands among NP patients, particularly within the core \\\"pain matrix\\\" network involving prefrontal-anterior cingulate-insula connections. It is worth noting that the mega analysis by Bott et al. \\u003csup\\u003e[\\u003cspan citationid=\\\"CR19\\\" class=\\\"CitationRef\\\"\\u003e19\\u003c/span\\u003e]\\u003c/sup\\u003ealso revealed the most robust association between large-scale brain network connections in the theta band (especially those involving the limbic system) and pain intensity. The study by Li et al\\u003csup\\u003e[\\u003cspan citationid=\\\"CR20\\\" class=\\\"CitationRef\\\"\\u003e20\\u003c/span\\u003e]\\u003c/sup\\u003e. demonstrated that persistent pain exacerbates individuals' difficulty in disengaging attention from pain-related words and was accompanied by heightened sensitivity of early delta/theta oscillations to \\\"expectancy violations.\\\" This aligns closely with our observations of hyperactive theta-band network activity and enhanced theta-gamma coupling.This widespread connectivity enhancement suggests that the NP brain exists in a highly synchronized yet potentially inefficient \\\"hyperconnected\\\" state, possibly disrupting the normal balance between network segregation and integration.\\u003c/p\\u003e\\u003cp\\u003eOf particular significance is our discovery of enhanced cross-frequency coupling (PAC) between occipital theta and prefrontal gamma activity in NP patients.Theta-gamma coupling represents a fundamental mechanism for neural information encoding and transmission\\u003csup\\u003e[\\u003cspan citationid=\\\"CR21\\\" class=\\\"CitationRef\\\"\\u003e21\\u003c/span\\u003e, \\u003cspan citationid=\\\"CR22\\\" class=\\\"CitationRef\\\"\\u003e22\\u003c/span\\u003e]\\u003c/sup\\u003e, where low-frequency theta oscillations modulate the timing of high-frequency gamma activity, thereby supporting working memory, attention, and long-range communication\\u003csup\\u003e[\\u003cspan citationid=\\\"CR23\\\" class=\\\"CitationRef\\\"\\u003e23\\u003c/span\\u003e, \\u003cspan citationid=\\\"CR24\\\" class=\\\"CitationRef\\\"\\u003e24\\u003c/span\\u003e]\\u003c/sup\\u003e.In NP, this enhanced coupling may indicate abnormally increased efficiency in information transmission and binding from sensory integration regions to higher-order cognitive regions in the prefrontal cortex, potentially forming the neural basis for pain memory consolidation, attentional capture, and cognitive bias. Contemporary research has begun to explore the therapeutic potential of modulating these cross-frequency interactions\\u003csup\\u003e[\\u003cspan citationid=\\\"CR25\\\" class=\\\"CitationRef\\\"\\u003e25\\u003c/span\\u003e, \\u003cspan citationid=\\\"CR26\\\" class=\\\"CitationRef\\\"\\u003e26\\u003c/span\\u003e]\\u003c/sup\\u003e .\\u003c/p\\u003e\\u003c/div\\u003e\\u003cdiv id=\\\"Sec12\\\" class=\\\"Section2\\\"\\u003e\\u003ch2\\u003e4.3 A New Paradigm in Pain Assessment: From Subjective Reporting to Objective Brain Network Decoding\\u003c/h2\\u003e\\u003cp\\u003eThe Graph-Generative Network (GGN) introduced in this study represents a paradigm shift in pain neuroimaging, moving beyond correlation-based biomarker discovery to establishing a causal-computational framework for objective pain phenotyping. Our model achieves transformative performance in both pain classification (98% accuracy) and intensity prediction, fundamentally redefining what is possible in pain assessment. Critically, GGN transcends the limitations of traditional spectral-feature approaches by directly learning the dynamic spatiotemporal architecture of pain\\u0026mdash;capturing how distributed brain networks collectively encode the pain experience in real time\\u003csup\\u003e[\\u003cspan additionalcitationids=\\\"CR28\\\" citationid=\\\"CR27\\\" class=\\\"CitationRef\\\"\\u003e27\\u003c/span\\u003e\\u0026ndash;\\u003cspan citationid=\\\"CR29\\\" class=\\\"CitationRef\\\"\\u003e29\\u003c/span\\u003e]\\u003c/sup\\u003e.\\u003c/p\\u003e\\u003cp\\u003eWhat distinguishes GGN is its capacity to generate dynamic, interpretable connectivity maps that reveal the precise neural circuits governing different pain states. These maps identify distinct network signatures\\u0026mdash;particularly in parietal-prefrontal and temporo-occipital pathways\\u0026mdash;that serve as computational biomarkers for pain subtypes. This represents the first demonstration that subjective pain experiences can be objectively decoded from spontaneous brain activity alone, without provocative stimulation or patient self-report.\\u003c/p\\u003e\\u003cp\\u003eThis work positions GGN as more than an analytical tool; it establishes a new paradigm where brain network dynamics become the primary language for understanding, measuring, and treating pain\\u0026mdash;finally liberating the field from its centuries-long reliance on subjective reporting and moving us toward a future where pain is understood through the intrinsic organization of brain networks.\\u003c/p\\u003e\\u003c/div\\u003e\\u003cdiv id=\\\"Sec13\\\" class=\\\"Section2\\\"\\u003e\\u003ch2\\u003e4.4 Research Limitations and Clinical Translation Prospects\\u003c/h2\\u003e\\u003cp\\u003eWhile this study demonstrates significant findings, several limitations warrant consideration. First, the cross-sectional design cannot establish causality, as these neural features could represent either causes, consequences, or both aspects of pain. Second, although non-NP pain control groups were included, potential confounding effects from neuropsychiatric comorbidities (such as anxiety and depression) require further investigation. Additionally, while the sample size is substantial for HD-EEG research, multicenter validation remains necessary to enhance generalizability. Recent methodological reviews have emphasized the importance of addressing these limitations in neuropathic pain research.\\u003c/p\\u003e\\u003cp\\u003eFrom a clinical translation perspective, despite current technical complexities in HD-EEG and GGN analysis, rapid advancements in neural technologies are lowering application barriers\\u003csup\\u003e[\\u003cspan citationid=\\\"CR7\\\" class=\\\"CitationRef\\\"\\u003e7\\u003c/span\\u003e, \\u003cspan citationid=\\\"CR30\\\" class=\\\"CitationRef\\\"\\u003e30\\u003c/span\\u003e, \\u003cspan citationid=\\\"CR31\\\" class=\\\"CitationRef\\\"\\u003e31\\u003c/span\\u003e]\\u003c/sup\\u003e. Portable EEG devices and cloud computing platforms are making EEG-based pain assessment tools increasingly feasible for clinical practice within the next 5\\u0026ndash;10 years \\u003csup\\u003e[\\u003cspan citationid=\\\"CR32\\\" class=\\\"CitationRef\\\"\\u003e32\\u003c/span\\u003e, \\u003cspan citationid=\\\"CR33\\\" class=\\\"CitationRef\\\"\\u003e33\\u003c/span\\u003e]\\u003c/sup\\u003e, particularly for assessment and treatment guidance in refractory pain patients\\u003csup\\u003e[\\u003cspan citationid=\\\"CR34\\\" class=\\\"CitationRef\\\"\\u003e34\\u003c/span\\u003e]\\u003c/sup\\u003e.\\u003csup\\u003e[\\u003cspan citationid=\\\"CR35\\\" class=\\\"CitationRef\\\"\\u003e35\\u003c/span\\u003e]\\u003c/sup\\u003eRecent technological developments have significantly improved the practicality of EEG-based pain assessment in clinical settings \\u003csup\\u003e[\\u003cspan citationid=\\\"CR7\\\" class=\\\"CitationRef\\\"\\u003e7\\u003c/span\\u003e, \\u003cspan citationid=\\\"CR36\\\" class=\\\"CitationRef\\\"\\u003e36\\u003c/span\\u003e]\\u003c/sup\\u003e.\\u003c/p\\u003e\\u003c/div\\u003e\"},{\"header\":\"Conclusion\",\"content\":\"\\u003cp\\u003eThis study reveals central mechanism characteristics of NP from both neural oscillation and network connectivity perspectives, establishing a precise identification model based on GGN. These findings not only provide new neural markers for objective assessment of NP but also advance our conceptualization of NP as a brain network disorder. Furthermore, the multidimensional analytical framework and computational methods proposed in this study provide a theoretical foundation for developing targeted neuromodulation therapies, potentially advancing pain medicine toward precision medicine approaches.\\u003c/p\\u003e\"},{\"header\":\"References\",\"content\":\"\\u003col\\u003e\\n\\u003cli\\u003eDavis KD,Aghaeepour N,Ahn AH,et al.Discovery and validation of biomarkers to aid the development of safe and effective pain therapeutics: challenges and opportunities. Nature Reviews Neurology, 2020. \\u003cstrong\\u003e16\\u003c/strong\\u003e(7): p. 381-400\\u003c/li\\u003e\\n\\u003cli\\u003eKovačević I,Pavić J,Filipović B,et al.Integrated approach to chronic pain\\u0026mdash;the role of psychosocial factors and multidisciplinary treatment: a narrative review. International journal of environmental research and public health, 2024. \\u003cstrong\\u003e21\\u003c/strong\\u003e(9): p. 1135\\u003c/li\\u003e\\n\\u003cli\\u003eSchultz DM,Bakke CH,Ruble HL,et al.Intrathecal drug delivery for intractable pain: identified patient satisfaction survey study comparing intrathecal dose with satisfaction, pain relief, and side effects. Neuromodulation: Technology at the Neural Interface, 2025\\u003c/li\\u003e\\n\\u003cli\\u003eLeroux A,Rzasa-Lynn R,Crainiceanu C,et al.Wearable devices: current status and opportunities in pain assessment and management. Digital Biomarkers, 2021. \\u003cstrong\\u003e5\\u003c/strong\\u003e(1): p. 89-102\\u003c/li\\u003e\\n\\u003cli\\u003eJi Y,Ma BJ,Guo XQ,et al.[Discussion on the composition and implementation of diagnosis and treatment strategies for whole field pain management strategy]. Zhonghua Yi Xue Za Zhi, 2023. \\u003cstrong\\u003e103\\u003c/strong\\u003e(39): p. 3083-3087.doi:10.3760/cma.j.cn112137-20230704-01135\\u003c/li\\u003e\\n\\u003cli\\u003ePatel M,Hasoon J,Diez Tafur R,et al.The Impact of Chronic Pain on Cognitive Function. Brain Sciences, 2025. \\u003cstrong\\u003e15\\u003c/strong\\u003e(6): p. 559\\u003c/li\\u003e\\n\\u003cli\\u003eAhmad B and Barkana BD.Pain and the Brain: A Systematic Review of Methods, EEG Biomarkers, Limitations, and Future Directions. Neurology International, 2025. \\u003cstrong\\u003e17\\u003c/strong\\u003e(4): p. 46\\u003c/li\\u003e\\n\\u003cli\\u003eMathew J,Perez TM,Adhia DB,et al.Is there a difference in EEG characteristics in acute, chronic, and experimentally induced musculoskeletal pain states? A systematic review. Clinical EEG and neuroscience, 2024. \\u003cstrong\\u003e55\\u003c/strong\\u003e(1): p. 101-120\\u003c/li\\u003e\\n\\u003cli\\u003eMay ES,Tiemann L,\\u0026Aacute;vila CG,et al.Assessing the predictive value of peak alpha frequency for the sensitivity to pain. Pain, 2022: p. 10.1097\\u003c/li\\u003e\\n\\u003cli\\u003eChowdhury NS,Bi C,Furman AJ,et al.Predicting individual pain sensitivity using a novel cortical biomarker signature. JAMA neurology, 2025. \\u003cstrong\\u003e82\\u003c/strong\\u003e(3): p. 237-246\\u003c/li\\u003e\\n\\u003cli\\u003eKleeva D,Soghoyan G,Biktimirov A,et al.Modulations in high-density EEG during the suppression of phantom-limb pain with neurostimulation in upper limb amputees. Cerebral Cortex, 2024. \\u003cstrong\\u003e34\\u003c/strong\\u003e(2): p. bhad504\\u003c/li\\u003e\\n\\u003cli\\u003eKannan S,Patel K,Di Basilio D,et al.Shared neural signatures in Functional Neurological Disorder and Chronic Pain: a multimodal narrative review. BMJ Neurology Open, 2025. \\u003cstrong\\u003e7\\u003c/strong\\u003e(2): p. e001032\\u003c/li\\u003e\\n\\u003cli\\u003eCavaleri R,McLain NJ,Heindel M,et al.Peak alpha frequency is related to the degree of widespread pain, but not pain intensity or duration, among people with urologic chronic pelvic pain syndrome. Pain Reports, 2025. \\u003cstrong\\u003e10\\u003c/strong\\u003e(2): p. e1251\\u003c/li\\u003e\\n\\u003cli\\u003eLi P,Yokoyama M,Okamoto D,et al.Resting-state EEG features modulated by depressive state in healthy individuals: insights from theta PSD, theta-beta ratio, frontal-parietal PLV, and sLORETA. Frontiers in Human Neuroscience, 2024. \\u003cstrong\\u003e18\\u003c/strong\\u003e: p. 1384330\\u003c/li\\u003e\\n\\u003cli\\u003eLi M,She K,Zhu P,et al.Chronic pain and comorbid emotional disorders: neural circuitry and neuroimmunity pathways. International Journal of Molecular Sciences, 2025. \\u003cstrong\\u003e26\\u003c/strong\\u003e(2): p. 436\\u003c/li\\u003e\\n\\u003cli\\u003eChmiel J and Kurpas D.Neural Correlates of Borderline Personality Disorder (BPD) Based on Electroencephalogram (EEG)\\u0026mdash;A Mechanistic Review. International Journal of Molecular Sciences, 2025. \\u003cstrong\\u003e26\\u003c/strong\\u003e(17): p. 8230\\u003c/li\\u003e\\n\\u003cli\\u003eHan S,Wang J,Zhang W,et al.Chronic pain\\u0026ndash;related cognitive deficits: preclinical insights into molecular, cellular, and circuit mechanisms. Molecular Neurobiology, 2024. \\u003cstrong\\u003e61\\u003c/strong\\u003e(10): p. 8123-8143\\u003c/li\\u003e\\n\\u003cli\\u003eManwatkar S,Puri M,Kumar B,et al.Tracing The Evolution Of Neuropathic Pain Markers: A Journey From Past Discoveries To Emerging Innovations And Future Directions. REDVET-Revista electr\\u0026oacute;nica de Veterinaria. \\u003cstrong\\u003e25\\u003c/strong\\u003e(1): p. 2024\\u003c/li\\u003e\\n\\u003cli\\u003eBott FS,Zebhauser PT,Hohn VD,et al.Exploring electroencephalographic chronic pain biomarkers: a mega-analysis. EBioMedicine, 2025. \\u003cstrong\\u003e120\\u003c/strong\\u003e: p. 105955.doi:10.1016/j.ebiom.2025.105955\\u003c/li\\u003e\\n\\u003cli\\u003eLi J,Lyu X,Li X,et al.Pain in focus: How persistent pain disrupts the attentional bias towards pain-related information. Neuroimage, 2025. \\u003cstrong\\u003e321\\u003c/strong\\u003e: p. 121539.doi:10.1016/j.neuroimage.2025.121539\\u003c/li\\u003e\\n\\u003cli\\u003eUrsino M and Pirazzini G.Theta\\u0026ndash;gamma coupling as a ubiquitous brain mechanism: Implications for memory, attention, dreaming, imagination, and consciousness. Current Opinion in Behavioral Sciences, 2024. \\u003cstrong\\u003e59\\u003c/strong\\u003e: p. 101433\\u003c/li\\u003e\\n\\u003cli\\u003ePirazzini G and Ursino M.Modeling the contribution of theta-gamma coupling to sequential memory, imagination, and dreaming. Frontiers in Neural Circuits, 2024. \\u003cstrong\\u003e18\\u003c/strong\\u003e: p. 1326609\\u003c/li\\u003e\\n\\u003cli\\u003eAi H,Zhang S,Si C,et al.Impaired theta and low-gamma directed information flow in the hippocampal-prefrontal circuit underlies working memory deficits in APP/PS1 mice. Behavioral and Brain Functions, 2025. \\u003cstrong\\u003e21\\u003c/strong\\u003e(1): p. 21\\u003c/li\\u003e\\n\\u003cli\\u003ePalacino F,Manganotti P and Benussi A.Targeting neural oscillations for cognitive enhancement in Alzheimer\\u0026rsquo;s disease. Medicina, 2025. \\u003cstrong\\u003e61\\u003c/strong\\u003e(3): p. 547\\u003c/li\\u003e\\n\\u003cli\\u003eGuo Z,Lin J-P,Simeone O,et al.Cross-frequency cortex\\u0026ndash;muscle interactions are abnormal in young people with dystonia. Brain Communications, 2024. \\u003cstrong\\u003e6\\u003c/strong\\u003e(2): p. fcae061\\u003c/li\\u003e\\n\\u003cli\\u003eMark JI,Riddle J,Gangwani R,et al.Cross-frequency coupling as a biomarker for early stroke recovery. Neurorehabilitation and neural repair, 2024. \\u003cstrong\\u003e38\\u003c/strong\\u003e(7): p. 506-517\\u003c/li\\u003e\\n\\u003cli\\u003eMohammadi H and Karwowski W.Graph neural networks in brain connectivity studies: Methods, challenges, and future directions. Brain Sciences, 2024. \\u003cstrong\\u003e15\\u003c/strong\\u003e(1): p. 17\\u003c/li\\u003e\\n\\u003cli\\u003eYang H,Huang R,Ye S,et al.A Review of Graph Neural Networks for Brain Diseases Analysis. Available at SSRN 5400612, \\u003c/li\\u003e\\n\\u003cli\\u003eZhang W and Hong Q.Modeling brain functional networks using graph neural networks: A review and clinical application. ICCK Transactions on Intelligent Systematics, 2024. \\u003cstrong\\u003e1\\u003c/strong\\u003e(2): p. 58-68\\u003c/li\\u003e\\n\\u003cli\\u003eParsa M,Rad HY,Vaezi H,et al.EEG-based classification of individuals with neuropsychiatric disorders using deep neural networks: A systematic review of current status and future directions. Comput Methods Programs Biomed, 2023. \\u003cstrong\\u003e240\\u003c/strong\\u003e: p. 107683.doi:10.1016/j.cmpb.2023.107683\\u003c/li\\u003e\\n\\u003cli\\u003ePu L,Lion KM,Todorovic M,et al.Portable EEG monitoring for older adults with dementia and chronic pain-a feasibility study. Geriatric Nursing, 2021. \\u003cstrong\\u003e42\\u003c/strong\\u003e(1): p. 124-128\\u003c/li\\u003e\\n\\u003cli\\u003eRice DA,Ozolins C,Biswas R,et al.Home-based EEG neurofeedback for the treatment of chronic pain: a randomized controlled clinical trial. The Journal of Pain, 2024. \\u003cstrong\\u003e25\\u003c/strong\\u003e(11): p. 104651\\u003c/li\\u003e\\n\\u003cli\\u003eTeel EF,Ocay DD,Blain-Moraes S,et al.Accurate classification of pain experiences using wearable electroencephalography in adolescents with and without chronic musculoskeletal pain. Frontiers in Pain Research, 2022. \\u003cstrong\\u003e3\\u003c/strong\\u003e: p. 991793\\u003c/li\\u003e\\n\\u003cli\\u003eLovelace JA,Miller J,Jacobs S,et al.Development of Machine Learning Algorithms Using EEG Data to Detect the Presence of Chronic Pain. medRxiv, 2024: p. 2024.2009. 2018.24313903\\u003c/li\\u003e\\n\\u003cli\\u003eChang JL,Nguyen P,Ruan QZ,et al.The Potential of Wearable, Modular Devices in Monitoring Functional Clinical Metrics in Patients Suffering from Chronic Pain. Current Pain and Headache Reports, 2025. \\u003cstrong\\u003e29\\u003c/strong\\u003e(1): p. 46\\u003c/li\\u003e\\n\\u003cli\\u003eAlshehri H,Al-Nafjan A and Aldayel M.Decoding Pain: A Comprehensive Review of Computational Intelligence Methods in Electroencephalography-Based Brain\\u0026ndash;Computer Interfaces. Diagnostics, 2025. \\u003cstrong\\u003e15\\u003c/strong\\u003e(3): p. 300\\u003c/li\\u003e\\n\\u003c/ol\\u003e\"}],\"fulltextSource\":\"\",\"fullText\":\"\",\"funders\":[],\"hasAdminPriorityOnWorkflow\":false,\"hasManuscriptDocX\":true,\"hasOptedInToPreprint\":true,\"hasPassedJournalQc\":\"\",\"hasAnyPriority\":true,\"hideJournal\":false,\"highlight\":\"\",\"institution\":\"\",\"isAcceptedByJournal\":false,\"isAuthorSuppliedPdf\":false,\"isDeskRejected\":\"\",\"isHiddenFromSearch\":false,\"isInQc\":false,\"isInWorkflow\":false,\"isPdf\":false,\"isPdfUpToDate\":true,\"isWithdrawnOrRetracted\":false,\"journal\":{\"display\":true,\"email\":\"info@researchsquare.com\",\"identity\":\"nature-portfolio\",\"isNatureJournal\":true,\"hasQc\":false,\"allowDirectSubmit\":false,\"externalIdentity\":\"\",\"sideBox\":\"\",\"snPcode\":\"\",\"submissionUrl\":\"\",\"title\":\"Nature Portfolio\",\"twitterHandle\":\"\",\"acdcEnabled\":false,\"dfaEnabled\":false,\"editorialSystem\":\"ejp\",\"reportingPortfolio\":\"\",\"inReviewEnabled\":true,\"inReviewRevisionsEnabled\":false},\"keywords\":\"\",\"lastPublishedDoi\":\"10.21203/rs.3.rs-8008823/v1\",\"lastPublishedDoiUrl\":\"https://doi.org/10.21203/rs.3.rs-8008823/v1\",\"license\":{\"name\":\"CC BY 4.0\",\"url\":\"https://creativecommons.org/licenses/by/4.0/\"},\"manuscriptAbstract\":\"\\u003cp\\u003eThe absence of objective, biologically-based diagnostics for neuropathic pain (NP) is a fundamental barrier to precision medicine in neurology and psychiatry. Reliance on subjective self-report leads to misdiagnosis, heterogeneous patient cohorts in clinical trials, and an inability to objectively evaluate therapeutic efficacy.\\u003c/p\\u003e\\n\\u003cp\\u003eTo address this, we leveraged the China Chronic Pain Cohort (CPCC) and high-density electroencephalography (HD-EEG) to identify objective neural signatures of NP. Our study enrolled 286 participants, including 117 NP patients, 56 non-neuropathic pain patients, and 113 healthy controls. HD-EEG analysis revealed distinct spatial functional connectivity patterns and neuronal oscillation coupling mechanisms characteristic of different pain categories.\\u003c/p\\u003e\\n\\u003cp\\u003eMultimodal analysis of HD-EEG revealed a distinct and reproducible neurophysiological signature of NP, characterized by aberrant oscillatory power and functional connectivity within a distributed network encompassing prefrontal, cingulate, and insular regions. Critically, we developed a novel Graph-Generative Network (GGN) model that translated these complex neural patterns into a clinically actionable tool. Our model differentiated NP from other pain types and healthy controls with exceptional accuracy (98%), and reliably predicted subjective pain intensity (VAS/NRS scores), effectively decoding a subjective state into an objective metric.\\u003c/p\\u003e\\n\\u003cp\\u003eThese findings suggest that HD-EEG signatures can serve as objective biomarkers for NP, offering new avenues for objective pain assessment and personalized treatment strategies. Future research should focus on validating these biomarkers in larger cohorts and exploring their potential for broader clinical application.\\u003c/p\\u003e\",\"manuscriptTitle\":\"High-Density Electroencephalographic Analysis of Neuropathic Pain - A Prospective, Multicenter Clinical Study\",\"msid\":\"\",\"msnumber\":\"\",\"nonDraftVersions\":[{\"code\":1,\"date\":\"2025-12-03 11:36:03\",\"doi\":\"10.21203/rs.3.rs-8008823/v1\",\"editorialEvents\":[],\"status\":\"published\",\"journal\":{\"display\":true,\"email\":\"info@researchsquare.com\",\"identity\":\"nature-communications\",\"isNatureJournal\":true,\"hasQc\":false,\"allowDirectSubmit\":false,\"externalIdentity\":\"NCOMMS\",\"sideBox\":\"Learn more about [Nature Communications](http://www.nature.com/ncomms/)\",\"snPcode\":\"\",\"submissionUrl\":\"https://mts-ncomms.nature.com/\",\"title\":\"Nature Communications\",\"twitterHandle\":\"\",\"acdcEnabled\":true,\"dfaEnabled\":true,\"editorialSystem\":\"ejp\",\"reportingPortfolio\":\"Nature Communications\",\"inReviewEnabled\":true,\"inReviewRevisionsEnabled\":false}}],\"origin\":\"\",\"ownerIdentity\":\"73545da1-ccae-4d5d-bc62-bee3b17f1bfd\",\"owner\":[],\"postedDate\":\"December 3rd, 2025\",\"published\":true,\"recentEditorialEvents\":[],\"rejectedJournal\":[],\"revision\":\"\",\"amendment\":\"\",\"status\":\"under-review\",\"subjectAreas\":[{\"id\":57429179,\"name\":\"Biological sciences/Neuroscience/Peripheral nervous system/Somatic system\"},{\"id\":57429180,\"name\":\"Health sciences/Biomarkers/Predictive markers\"}],\"tags\":[],\"updatedAt\":\"2025-12-03T11:36:03+00:00\",\"versionOfRecord\":[],\"versionCreatedAt\":\"2025-12-03 11:36:03\",\"video\":\"\",\"vorDoi\":\"\",\"vorDoiUrl\":\"\",\"workflowStages\":[]},\"version\":\"v1\",\"identity\":\"rs-8008823\",\"journalConfig\":\"researchsquare\"},\"__N_SSP\":true},\"page\":\"/article/[identity]/[[...version]]\",\"query\":{\"redirect\":\"/article/rs-8008823\",\"identity\":\"rs-8008823\",\"version\":[\"v1\"]},\"buildId\":\"8U1c8b4HqxoKbykW_rLl7\",\"isFallback\":false,\"isExperimentalCompile\":false,\"dynamicIds\":[84888],\"gssp\":true,\"scriptLoader\":[]}","source_license":"CC-BY-4.0","license_restricted":false}