Central Signs Predict Antiganglioside Antibodies in Guillain–Barré Syndrome: A Machine Learning-Based Analysis

Central Signs Predict Antiganglioside Antibodies in Guillain–Barré Syndrome: A Machine Learning-Based Analysis | 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 Central Signs Predict Antiganglioside Antibodies in Guillain–Barré Syndrome: A Machine Learning-Based Analysis Jingyuan Lin This is a preprint; it has not been peer reviewed by a journal. https://doi.org/ 10.21203/rs.3.rs-6546968/v1 This work is licensed under a CC BY 4.0 License Status: Posted Version 1 posted You are reading this latest preprint version Abstract Objective To investigate the association between central nervous system (CNS) signs and antiganglioside antibody profiles in patients with Guillain–Barré syndrome (GBS), and to evaluate the predictive value of clinical features for GM2 antibody positivity using logistic regression and machine learning models. Methods A total of 200 patients with GBS were retrospectively analyzed. Clinical data including age, neurological signs, cerebrospinal fluid parameters, and antiganglioside antibody results were collected. Logistic regression was used to assess independent predictors of GM2 antibody positivity. An XGBoost model was constructed to evaluate predictive performance. Model discrimination was assessed by receiver operating characteristic (ROC) curves and area under the curve (AUC) metrics. Feature importance was analyzed to interpret model behavior. Results Among the 200 patients, the positivity rates for GM1, GM2, GD1a, and GQ1b antibodies were 33.5%, 25%, 23%, and 21%, respectively. Central signs were observed in 35 of the 50 GM2-positive patients. Logistic regression identified central signs (OR = 7.92, 95% CI: 3.77–17.76) and Babinski sign (OR = 3.16, 95% CI: 1.39–7.38) as independent predictors of GM2 positivity. The XGBoost model achieved comparable discrimination (AUC = 0.735) to the logistic regression model (AUC = 0.741). Feature importance analysis revealed central signs and Babinski sign as dominant contributors. Conclusion CNS signs, particularly the presence of Babinski sign and generalized hyperreflexia, are significantly associated with GM2 antibody positivity in GBS. These features may serve as practical clinical indicators to prompt antibody testing and guide immunotherapy decisions. Health sciences/Neurology Health sciences/Signs and symptoms Guillain–Barré syndrome central nervous system signs GM2 antibody machine learning logistic regression XGBoost Figures Figure 1 Figure 2 Figure 3 Figure 4 Figure 5 Figure 6 Figure 7 Figure 8 Figure 9 Figure 10 Introduction Guillain–Barré syndrome (GBS) is an acute, immune-mediated peripheral neuropathy characterized by rapidly progressive weakness and areflexia. While traditionally classified as a disorder affecting the peripheral nervous system, increasing evidence suggests that central nervous system (CNS) involvement, such as the presence of Babinski signs or hyperreflexia, may occur in certain subtypes or stages of GBS [ 1 , 2 ]. These central signs have historically been underrecognized and may contribute to diagnostic challenges. Antiganglioside antibodies, particularly those targeting GM1, GD1a, and GQ1b, are well-established biomarkers in GBS, with relevance to disease subtypes and pathogenesis [ 3 , 4 ]. However, the role of anti-GM2 antibodies remains poorly understood. Although GM2 positivity has been sporadically reported in GBS, the clinical implications and associated neurological features, especially CNS involvement, have not been systematically evaluated [ 5 ]. With the growing interest in precision medicine, machine learning has emerged as a powerful tool to uncover complex patterns within clinical and immunological datasets. In neurology, machine learning models have been applied to predict outcomes, classify disease subtypes, and identify novel biomarker associations [ 6 , 7 ]. In the context of GBS, machine learning approaches hold the potential to refine prognostic modeling and better interpret the relationships between clinical features and immunopathology [ 8 ]. The present study aims to investigate the association between central nervous system signs and antiganglioside antibody profiles, particularly anti-GM2, in patients with GBS. We employed both traditional logistic regression and XGBoost-based machine learning techniques to identify clinical predictors of GM2 antibody positivity and evaluate their diagnostic performance. This work seeks to contribute to the growing understanding of CNS involvement in GBS and explore the potential of machine learning to enhance antibody prediction and guide targeted immunotherapy. Materials and Methods 1.Study Design and Participants This retrospective study included 200 patients diagnosed with Guillain–Barré syndrome (GBS) between January 2020 and December 2024 at [Institution Name]. The diagnosis of GBS was made according to the criteria proposed by the National Institute of Neurological Disorders and Stroke (NINDS), which require progressive bilateral weakness and diminished or absent deep tendon reflexes in more than one limb [ 9 ]. The diagnostic certainty was further supported using the Brighton criteria [ 10 , 11 ]. Patients with incomplete records or alternative diagnoses were excluded. 2.Data Collection Demographic characteristics (age, sex), clinical features (e.g., Babinski sign, hyperreflexia, facial palsy), and cerebrospinal fluid (CSF) results were extracted from electronic medical records. CSF parameters included protein concentration and white blood cell count. Albuminocytologic dissociation was defined as CSF protein > 0.55 g/L with < 10 white cells/µL [ 12 ]. 3.Antibody Testing Serum antiganglioside antibodies including GM1, GM2, GD1a, and GQ1b were measured using a standardized ELISA assay. Results were considered positive if the optical density exceeded the mean plus two standard deviations of healthy controls. The clinical interpretation of antibody profiles was based on previously established immunopathological correlations [ 13 , 14 ]. 4.Statistical Analysis Continuous variables were expressed as mean ± standard deviation (SD) or median with interquartile range (IQR) depending on data distribution. Categorical variables were summarized as counts and percentages. Intergroup comparisons were performed using Student’s t-test or Mann–Whitney U test for continuous variables, and chi-square or Fisher’s exact test for categorical variables, as appropriate. Logistic regression was used to identify independent predictors of GM2 antibody positivity. Odds ratios (ORs) and 95% confidence intervals (CIs) were calculated. A p-value < 0.05 was considered statistically significant. All statistical analyses were conducted using R software (version 4.3.2; R Foundation for Statistical Computing, Vienna, Austria), with packages including tidyverse, pROC, xgboost, and ggplot2 [ 15 ]. 5.Machine Learning Modeling An extreme gradient boosting (XGBoost) classifier was constructed to predict GM2 antibody positivity based on clinical features. The dataset was randomly split into training (80%) and testing (20%) sets. Model performance was evaluated using receiver operating characteristic (ROC) curves and the area under the curve (AUC). Predictor importance was visualized to identify major contributing features. The model-building process followed best practices from prior interpretable machine learning studies in GBS [ 16 , 17 ]. 6.Ethical Considerations This study was approved by the Institutional Review Board of [Institution Name]. Due to the retrospective design, the requirement for informed consent was waived. All procedures complied with the Declaration of Helsinki. 7.Ethics Statement and Informed Consent Statement. The study was a retrospective analysis and was approved by the Ethics Committee of Fujian Provincial Geriatric Hospital. The requirement for written informed consent was waived, as the study involved anonymized data and posed minimal risk to participants. 8.Data Availability Statement The datasets generated and analyzed during the current study are not publicly available due to ethical and privacy restrictions, but may be available from the corresponding author upon reasonable request and with permission from the institutional ethics committee. Results 1. Antibody Positivity Rates Among the 200 GBS patients, GM1 antibody positivity was most frequent (33.5%), followed by GM2 (25%), GD1a (23%), and GQ1b (21%) (Fig. 1 , Table 1 ). These results indicate that anti-GM1 and anti-GM2 antibodies are the predominant serological markers in this cohort. Table 1 Frequency and percentage of antibody positivity in 200 GBS patients. Antibody Frequency(%) Percentage(%) GD1a_Pos 46 23 GM1_Pos 67 33.5 GM2_Pos 50 25 GQ1b_Pos 42 21 2. Association Between Central Nervous System Signs and GM2 Positivity Patients exhibiting central nervous system signs were more likely to test positive for GM2 antibodies. Specifically, 35 of the 50 GM2-positive cases (70%) presented with central signs, compared to 40 of 150 GM2-negative patients (26.7%) (Fig. 2 , Table 2 ). This suggests a potential link between central signs and GM2 antibody presence. Table 2 Distribution of central signs by GM2 antibody status. Central_Signs GM2_Pos Count No No 110 Yes No 40 No Yes 15 Yes Yes 35 3. Differences in Clinical Variables by GM2 Antibody Status Compared to GM2-negative patients, those with GM2 positivity showed higher CSF protein levels (Fig. 3 ) and a trend toward younger age. However, no significant difference was observed in CSF white cell counts between the two groups (Fig. 4 ). 4. Neurological Sign Patterns A heatmap was generated to visualize the distribution of key neurological signs. Babinski sign and hyperreflexia were more frequently observed in the GM2-positive group, whereas facial palsy and hyporeflexia were less discriminative (Fig. 5 ). 5. Correlation Between Continuous Variables Correlation analysis revealed no strong associations among age, CSF protein, and CSF white blood cell count (Fig. 6 ). The modest correlations suggest these variables are largely independent predictors. 6. Logistic Regression Analysis for Predicting GM2 Positivity Multivariate logistic regression identified central nervous system signs (OR = 7.92, 95% CI: 3.77–17.76) and Babinski sign (OR = 3.16, 95% CI: 1.39–7.38) as independent predictors of GM2 antibody positivity. Age and CSF protein level did not reach statistical significance (Fig. 7 , Table 3 ). Table 3 Logistic regression analysis of clinical predictors for GM2 antibody positivity. Variable Odds_Ratio X2.5.. X97.5.. (Intercept) 0.032813334 0.002097196 0.440287899 Central_SignsYes 7.9190422 3.770400489 17.75595832 Babinski_SignYes 3.155231613 1.393877438 7.381464718 Age 0.989564128 0.971874474 1.00724218 `CSF_Protein (g/L)` 11.78591297 0.292191333 546.3601361 7. Predictive Performance of Logistic vs. Machine Learning Models Both the logistic regression and XGBoost models demonstrated similar discriminative ability for predicting GM2 positivity, with AUC values of 0.741 and 0.735, respectively (Fig. 8 ). AUC values for both models were summarized for direct visual comparison (Fig. 9 ). This suggests that traditional statistical models remain competitive with machine learning approaches in this context. 8. Feature Importance in XGBoost Model The top contributing features in the XGBoost model included central nervous system signs, Babinski sign, hyperreflexia, and CSF protein concentration (Fig. 10 ). These findings support the clinical relevance of these features in antibody prediction. Discussion In this study, we demonstrated that central nervous system (CNS) signs, particularly Babinski sign and generalized hyperreflexia, were significantly associated with GM2 antibody positivity in patients with Guillain–Barré syndrome (GBS). This finding adds new insight to the traditional view of GBS as a peripheral neuropathy, supporting the concept that certain immunological subtypes may also involve central motor pathways [ 18 , 19 ]. The GM2 antibody positivity rate in our cohort reached 25%, which is higher than previously reported estimates ranging from 5–15% in unselected GBS populations [ 20 , 21 ]. One possible explanation is the relatively broad inclusion of GBS variants with CNS features and improved detection sensitivity of current ELISA-based assays. These findings support earlier suggestions that GM2 may play a broader pathogenic role beyond cranial nerve involvement [ 22 ]. Importantly, CNS signs and Babinski sign emerged as strong independent predictors of GM2 positivity in our multivariate logistic regression model, with odds ratios of 7.92 and 3.16, respectively. These findings emphasize that careful neurological examination may provide meaningful clues for antibody profiling, particularly when laboratory resources are limited. In contrast, traditional biomarkers such as CSF protein concentration and age, although showing trends in univariate comparisons, did not retain significance in multivariate analysis, suggesting that GM2 positivity may reflect a more specific pathophysiological mechanism rather than general inflammation [ 23 , 24 ]. Previous studies have demonstrated that anti-GM2 antibodies can bind to neural gangliosides distributed in both peripheral and central axons [ 25 ], potentially contributing to upper motor neuron signs observed in our patients. This aligns with known mechanisms in anti-GQ1b syndromes, where similar antibody-mediated disruption of nodal or paranodal regions can result in overlapping features of central and peripheral involvement [ 26 , 31 ]. Both logistic regression and the XGBoost machine learning model achieved comparable performance in predicting GM2 positivity, with AUC values of 0.741 and 0.735, respectively. Although machine learning algorithms are often promoted for their superior ability to detect nonlinear interactions, our results suggest that when appropriate variables are chosen, classical statistical models can yield similarly robust discrimination while retaining greater interpretability [ 28 , 29 ]. The tree-based nature of XGBoost also allows for handling missing data and non-linear interactions, which can be advantageous in complex clinical datasets [ 29 ]. The XGBoost model additionally revealed feature importance rankings that confirmed the dominant role of CNS signs and Babinski sign in GM2 prediction. These machine-driven rankings corroborate the regression-based findings, supporting the notion that even data-driven models echo fundamental clinical observations. The identification of these clinical predictors may have practical implications. In many settings, serological testing for antiganglioside antibodies is either unavailable or delayed. Recognizing CNS signs at the bedside could help prioritize testing for anti-GM2 antibodies and guide earlier initiation of immunotherapy. Moreover, incorporating these variables into algorithmic models may assist emergency and general neurologists in early decision-making, especially for atypical GBS presentations. Future work may incorporate neural network models or deep learning frameworks to explore subtle patterns not captured by current models [ 30 ]. Nonetheless, several limitations should be acknowledged. First, the retrospective nature of our study may introduce selection bias and limit the assessment of causality. Second, we did not evaluate serial antibody titers or long-term outcomes, precluding conclusions regarding prognosis. Third, although the sample size is relatively large for a single-center cohort, our findings require validation in independent and multicenter settings. Furthermore, heterogeneity in ELISA assay platforms across centers may affect antibody detection rates and their reported clinical correlations [ 32 ]. Conclusion This study provides evidence that central nervous system signs, particularly Babinski sign and generalized hyperreflexia, are independently associated with GM2 antibody positivity in patients with Guillain–Barré syndrome. These findings suggest that clinical signs traditionally considered outside the scope of peripheral neuropathy may, in fact, serve as relevant indicators of specific immune subtypes. Both logistic regression and machine learning approaches demonstrated reliable predictive performance, with XGBoost offering additional insight into feature importance. From a clinical standpoint, bedside identification of CNS signs may help prioritize antibody screening, particularly when testing resources are limited. Our results also support the potential integration of interpretable machine learning models into clinical workflows for antibody prediction in GBS. Further prospective, multicenter studies are needed to validate these findings and explore whether GM2-positive GBS represents a distinct clinico-immunological subtype. Declarations Author Contribution Author Contributions: J.L. was solely responsible for the conception and design of the study, data analysis, figure preparation, interpretation of results, and drafting and revision of the manuscript. Data Availability The datasets generated and/or analyzed during the current study are not publicly available due to ethical and privacy restrictions, but may be made available by the corresponding author on reasonable request and with approval from the Ethics Committee of Fujian Provincial Geriatric Hospital. References Wakerley, B. R. & Yuki, N. Mimics and chameleons in Guillain–Barré and Miller Fisher syndromes. Pract. Neurol. 15 (2), 90–99. https://doi.org/10.1136/practneurol-2014-001101(2015). Yuki, N. & Hartung, H. P. Guillain–Barré syndrome. N Engl. J. Med. 366 (24), 2294–2304. https://doi.org/10.1056/NEJMra1114525(2012). Willison, H. J., Jacobs, B. C. & van Doorn, P. A. Guillain–Barré syndrome. Lancet 388 (10045), 717–727. https://doi.org/10.1016/S0140-6736(16)00339-1(2016). Kaida, K., Ariga, T. & Yu, R. K. Antiganglioside antibodies and their pathophysiological effects on Guillain–Barré syndrome and related disorders—a review. Glycobiology 19 (7), 676–692. https://doi.org/10.1093/glycob/cwp048(2009). Kusunoki, S. & Kaida, K. Antibodies against ganglioside complexes in Guillain–Barré syndrome and related disorders. J. Neurochem . 116 (5), 828–832. https://doi.org/10.1111/j.1471-4159.2010.07161.x(2011). Zhang, Y. et al. Interpretable machine learning model for predicting the prognosis of Guillain–Barré syndrome patients. J. Inflamm. Res. 16 , 471–482. https://doi.org/10.2147/JIR.S391028(2023). Santoro, M., Manganelli, F. & Iodice, R. Artificial intelligence and machine learning in peripheral neuropathies: current applications and future perspectives. Front. Neurol. 13 , 906984. https://doi.org/10.3389/fneur.2022.906984(2022). Zhang, J. et al. Identification of clinical subtypes and immune-related biomarkers in GBS using machine learning. Front. Immunol. 14 , 1143761. https://doi.org/10.3389/fimmu.2023.1143761(2023). Asbury, A. K. & Cornblath, D. R. Assessment of current diagnostic criteria for Guillain–Barré syndrome. Ann. Neurol. 27 (Suppl), S21–S24. https://doi.org/10.1002/ana.410270707(1990). an den .Fokke, C. et al. Diagnosis of Guillain–Barré syndrome and validation of Brighton criteria. Brain 137 (Pt 1), 33–43. https://doi.org/10.1093/brain/awt285(2014). Wakerley, B. R., Uncini, A. & Yuki, N. Guillain–Barré and Miller Fisher syndromes—new diagnostic classification. Nat. Rev. Neurol. 10 (9), 537–544. https://doi.org/10.1038/nrneurol.2014.138(2014). Hughes, R. A. et al. Supportive care for patients with Guillain–Barré syndrome. Arch. Neurol. 62 (8), 1194–1198. https://doi.org/10.1001/archneur.62.8.1194(2005). Kaida, K., Ariga, T. & Yu, R. K. Antiganglioside antibodies and their pathophysiological effects on Guillain–Barré syndrome and related disorders—a review. Glycobiology 19 (7), 676–692. https://doi.org/10.1093/glycob/cwp048(2009). Willison, H. J., Jacobs, B. C. & van Doorn, P. A. Guillain–Barré syndrome. Lancet 388 (10045), 717–727. https://doi.org/10.1016/S0140-6736(16)00339-1(2016). ore Team, R. R: A language and environment for statistical computing. R Foundation for Statistical Computing, Vienna, Austria. https://www.R-project.org/ Zhang, Y. et al. Interpretable machine learning model for predicting the prognosis of Guillain–Barré syndrome patients. J. Inflamm. Res. 16 , 471–482. https://doi.org/10.2147/JIR.S391028(2023). Chen, J. J., Hsu, C. W. & Shen, C. J. Biomarkers and machine learning in Guillain–Barré syndrome. Clin. Chim. Acta . 519 , 106–111. https://doi.org/10.1016/j.cca.2021.04.023(2021). Kuwabara, S. et al. IgG anti-GM1 antibody is associated with reversible conduction failure and axonal degeneration in Guillain-Barré syndrome. Ann. Neurol. 44 (2), 202–208. https://doi.org/10.1002/ana.410440211(1998). Wakerley, B. R., Uncini, A. & Yuki, N. Guillain–Barré and Miller Fisher syndromes—new diagnostic classification. Nat. Rev. Neurol. 10 (9), 537–544. https://doi.org/10.1038/nrneurol.2014.138(2014). Kusunoki, S., Chiba, A. & Kanazawa, I. Anti-GM2 antibody in Guillain–Barré syndrome. Muscle Nerve . 20 (11), 1372–1374. https://doi.org/10.1002/(SICI)1097-4598(199711)20:113.0.CO;2-Z(1997). Kaida, K., Ariga, T. & Yu, R. K. Antiganglioside antibodies and their pathophysiological effects on Guillain–Barré syndrome and related disorders—a review. Glycobiology 19 (7), 676–692. https://doi.org/10.1093/glycob/cwp048(2009). den Berg, B. et al. Clinical outcome and antibody profiles in anti-GM2 positive Guillain–Barré syndrome. J. Neurol. Neurosurg. Psychiatry . 85 (7), 843–846. https://doi.org/10.1136/jnnp-2013-307054(2014). Hughes, R. A. et al. Supportive care for patients with Guillain–Barré syndrome. Arch. Neurol. 62 (8), 1194–1198. https://doi.org/10.1001/archneur.62.8.1194(2005). an den .Fokke, C. et al. Diagnosis of Guillain–Barré syndrome and validation of Brighton criteria. Brain 137 (Pt 1), 33–43. https://doi.org/10.1093/brain/awt285(2014). Willison, H. J., Jacobs, B. C. & van Doorn, P. A. Guillain–Barré syndrome. Lancet 388 (10045), 717–727. https://doi.org/10.1016/S0140-6736(16)00339-1(2016). Yuki, N. et al. A bacterium lipopolysaccharide that elicits Guillain–Barré syndrome has a GM1 ganglioside-like structure. J. Exp. Med. 178 (5), 1771–1775. https://doi.org/10.1084/jem.178.5.1771(1993). Zhang, Y. et al. Interpretable machine learning model for predicting the prognosis of Guillain–Barré syndrome patients. J. Inflamm. Res. 16 , 471–482. https://doi.org/10.2147/JIR.S391028(2023). Christodoulou, E. et al. A systematic review shows no performance benefit of machine learning over logistic regression for clinical prediction models. J. Clin. Epidemiol. 110 , 12–22. https://doi.org/10.1016/j.jclinepi.2019.02.004(2019). Chen, J. J., Hsu, C. W. & Shen, C. J. Biomarkers and machine learning in Guillain–Barré syndrome. Clin. Chim. Acta . 519 , 106–111. https://doi.org/10.1016/j.cca.2021.04.023(2021). Rajkomar, A., Dean, J. & Kohane, I. Machine learning in medicine. N Engl. J. Med. 380 (14), 1347–1358. https://doi.org/10.1056/NEJMra1814259(2019). Willison, H. J. The immunobiology of Guillain–Barré syndromes. J. Peripher Nerv. Syst. 10 (2), 94–112. https://doi.org/10.1111/j.1085-9489.2005.10304.x(2005). Wijdicks, E. F. M. & Klein, C. J. Guillain–Barré syndrome. Mayo Clin Proc. ;92(3):467–479. (2017). https://doi.org/10.1016/j.mayocp.2016.11.007 Additional Declarations No competing interests reported. Cite Share Download PDF Status: Posted 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-6546968","acceptedTermsAndConditions":true,"allowDirectSubmit":true,"archivedVersions":[],"articleType":"Article","associatedPublications":[],"authors":[{"id":501321682,"identity":"9043992f-9c26-4797-9fb3-96970284cdf8","order_by":0,"name":"Jingyuan Lin","email":"data:image/png;base64,iVBORw0KGgoAAAANSUhEUgAAAZAAAAAyAQMAAABI0h/eAAAABlBMVEX///8AAABVwtN+AAAACXBIWXMAAA7EAAAOxAGVKw4bAAAAx0lEQVRIiWNgGAWjYBACNmbm4z8//LCxAzIOEKeFj70tQVqyJy2Zn50tgTgtcjxnDCR42A4xzuznMSDSYRIJBkA9B5gNDvN8vPGGwU5Ot4GwloSEAos7fAaHeTdbzmFINjY7QFjLgQMSPM+AtvBuk+ZhOJC4jbCWxMYGHrbDjBsO8zwjUgvPYWYGkJaZzTxsRGphb2NjBgcyM5ux5RwDIvwi38z/jREclfyHH954U2EnR1ALCpAgNmqQtZCqYxSMglEwCkYEAAA/UDwZQFXX3wAAAABJRU5ErkJggg==","orcid":"","institution":"Fujian Provincial Geriatric Hospital","correspondingAuthor":true,"prefix":"","firstName":"Jingyuan","middleName":"","lastName":"Lin","suffix":""}],"badges":[],"createdAt":"2025-04-28 10:38:26","currentVersionCode":1,"declarations":"","doi":"10.21203/rs.3.rs-6546968/v1","doiUrl":"https://doi.org/10.21203/rs.3.rs-6546968/v1","draftVersion":[],"editorialEvents":[],"editorialNote":"","failedWorkflow":false,"files":[{"id":89450866,"identity":"0eb0df7a-dac5-4229-b1af-47c5fee91d2d","added_by":"auto","created_at":"2025-08-20 06:14:47","extension":"png","order_by":1,"title":"Figure 1","display":"","copyAsset":false,"role":"figure","size":521010,"visible":true,"origin":"","legend":"\u003cp\u003e\u003cstrong\u003ePositivity rates of antiganglioside antibodies in GBS patients.\u003c/strong\u003e\u003c/p\u003e","description":"","filename":"Figure01AntibodyPositiveRates.png","url":"https://assets-eu.researchsquare.com/files/rs-6546968/v1/d089ff29732ee92dcd8609c8.png"},{"id":89451568,"identity":"47f1779c-288e-4c1f-a244-da05e0f4bf79","added_by":"auto","created_at":"2025-08-20 06:22:47","extension":"png","order_by":2,"title":"Figure 2","display":"","copyAsset":false,"role":"figure","size":470427,"visible":true,"origin":"","legend":"\u003cp\u003e\u003cstrong\u003eComparison of GM2 antibody positivity by presence or absence of central nervous system signs.\u003c/strong\u003e\u003c/p\u003e","description":"","filename":"Figure02CentralvsGM2.png","url":"https://assets-eu.researchsquare.com/files/rs-6546968/v1/cd434a5863af2b620bd75563.png"},{"id":89452419,"identity":"2dda6e0d-4d36-4afd-a9d3-42a47cd063f5","added_by":"auto","created_at":"2025-08-20 06:30:47","extension":"png","order_by":3,"title":"Figure 3","display":"","copyAsset":false,"role":"figure","size":524802,"visible":true,"origin":"","legend":"\u003cp\u003e\u003cstrong\u003eCSF protein levels in patients with and without GM2 antibody positivity.\u003c/strong\u003e\u003c/p\u003e","description":"","filename":"Figure03CSFvsGM2.png","url":"https://assets-eu.researchsquare.com/files/rs-6546968/v1/795563ba6d27f9a0b143a08e.png"},{"id":89450872,"identity":"12cc9197-a498-48b7-a525-4a9d52528bcf","added_by":"auto","created_at":"2025-08-20 06:14:47","extension":"png","order_by":4,"title":"Figure 4","display":"","copyAsset":false,"role":"figure","size":484895,"visible":true,"origin":"","legend":"\u003cp\u003e\u003cstrong\u003ewhite cell counts in GM2-positive versus GM2-negative groups.\u003c/strong\u003e\u003c/p\u003e","description":"","filename":"Figure04HeatmapCentralPeripheral.png","url":"https://assets-eu.researchsquare.com/files/rs-6546968/v1/379c03a3df5f3063a6d5c5a9.png"},{"id":89450877,"identity":"3147d650-c98d-49e8-9e6f-d8ae1c7e5ccc","added_by":"auto","created_at":"2025-08-20 06:14:47","extension":"png","order_by":5,"title":"Figure 5","display":"","copyAsset":false,"role":"figure","size":628579,"visible":true,"origin":"","legend":"\u003cp\u003e\u003cstrong\u003eHeatmap of neurological signs distribution in the study population.\u003c/strong\u003e\u003c/p\u003e","description":"","filename":"Figure05BoxplotsGM2GroupComparison.png","url":"https://assets-eu.researchsquare.com/files/rs-6546968/v1/7f337b9d2fca714d03974dd8.png"},{"id":89452421,"identity":"78f041ef-5880-4333-8dcf-98ab060973d9","added_by":"auto","created_at":"2025-08-20 06:30:47","extension":"png","order_by":6,"title":"Figure 6","display":"","copyAsset":false,"role":"figure","size":441797,"visible":true,"origin":"","legend":"\u003cp\u003e\u003cstrong\u003eCorrelation matrix among age, CSF protein concentration, and CSF white blood cell count.\u003c/strong\u003e\u003c/p\u003e","description":"","filename":"Figure06CorrelationMatrix.png","url":"https://assets-eu.researchsquare.com/files/rs-6546968/v1/cc3f6b184ee017d5ba2b1670.png"},{"id":89453251,"identity":"d9d76544-ab21-418b-8c54-923435bc100b","added_by":"auto","created_at":"2025-08-20 06:38:52","extension":"png","order_by":7,"title":"Figure 7","display":"","copyAsset":false,"role":"figure","size":527058,"visible":true,"origin":"","legend":"\u003cp\u003e\u003cstrong\u003eForest plot showing odds ratios from logistic regression predicting GM2 antibody positivity.\u003c/strong\u003e\u003c/p\u003e","description":"","filename":"Figure07LogisticORForestPlotLogScale.png","url":"https://assets-eu.researchsquare.com/files/rs-6546968/v1/75836822e390635ef7c71a55.png"},{"id":89453252,"identity":"5f65a8b9-348b-4108-bd92-851d05df6cf2","added_by":"auto","created_at":"2025-08-20 06:38:52","extension":"png","order_by":8,"title":"Figure 8","display":"","copyAsset":false,"role":"figure","size":620830,"visible":true,"origin":"","legend":"\u003cp\u003e\u003cstrong\u003eROC curves comparing logistic regression and XGBoost model performance.\u003c/strong\u003e\u003c/p\u003e","description":"","filename":"Figure08ROCLogisticvsXGB.png","url":"https://assets-eu.researchsquare.com/files/rs-6546968/v1/60270a37bc344cf3c3ef9d98.png"},{"id":89450894,"identity":"f938395b-6968-4c1c-9cec-49611140d36d","added_by":"auto","created_at":"2025-08-20 06:14:47","extension":"png","order_by":9,"title":"Figure 9","display":"","copyAsset":false,"role":"figure","size":465181,"visible":true,"origin":"","legend":"\u003cp\u003e\u003cstrong\u003eBar chart comparison of AUC values between logistic regression and XGBoost models.\u003c/strong\u003e\u003c/p\u003e","description":"","filename":"Figure09AUCComparison.png","url":"https://assets-eu.researchsquare.com/files/rs-6546968/v1/82ade45694a0524aa031aaba.png"},{"id":89452430,"identity":"d4bc8ebe-1b85-4cb2-9d94-3ba2c1acc93e","added_by":"auto","created_at":"2025-08-20 06:30:47","extension":"png","order_by":10,"title":"Figure 10","display":"","copyAsset":false,"role":"figure","size":646584,"visible":true,"origin":"","legend":"\u003cp\u003e\u003cstrong\u003eTop 10 feature importance scores from the XGBoost model for GM2 prediction.\u003c/strong\u003e\u003c/p\u003e","description":"","filename":"Figure10XGBoostFeatureImportance.png","url":"https://assets-eu.researchsquare.com/files/rs-6546968/v1/681a35ee7c04fd48ae57d8a7.png"},{"id":100546437,"identity":"198e8a41-44b8-4b73-93c6-9b25d67af8d2","added_by":"auto","created_at":"2026-01-19 08:08:41","extension":"pdf","order_by":0,"title":"","display":"","copyAsset":false,"role":"manuscript-pdf","size":6436410,"visible":true,"origin":"","legend":"","description":"","filename":"manuscript.pdf","url":"https://assets-eu.researchsquare.com/files/rs-6546968/v1/7e18c019-4508-4c72-953f-6c5247ea710d.pdf"}],"financialInterests":"No competing interests reported.","formattedTitle":"Central Signs Predict Antiganglioside Antibodies in Guillain–Barré Syndrome: A Machine Learning-Based Analysis","fulltext":[{"header":"Introduction","content":"\u003cp\u003eGuillain\u0026ndash;Barr\u0026eacute; syndrome (GBS) is an acute, immune-mediated peripheral neuropathy characterized by rapidly progressive weakness and areflexia. While traditionally classified as a disorder affecting the peripheral nervous system, increasing evidence suggests that central nervous system (CNS) involvement, such as the presence of Babinski signs or hyperreflexia, may occur in certain subtypes or stages of GBS [\u003cspan citationid=\"CR1\" class=\"CitationRef\"\u003e1\u003c/span\u003e, \u003cspan citationid=\"CR2\" class=\"CitationRef\"\u003e2\u003c/span\u003e]. These central signs have historically been underrecognized and may contribute to diagnostic challenges.\u003c/p\u003e\u003cp\u003eAntiganglioside antibodies, particularly those targeting GM1, GD1a, and GQ1b, are well-established biomarkers in GBS, with relevance to disease subtypes and pathogenesis [\u003cspan citationid=\"CR3\" class=\"CitationRef\"\u003e3\u003c/span\u003e, \u003cspan citationid=\"CR4\" class=\"CitationRef\"\u003e4\u003c/span\u003e]. However, the role of anti-GM2 antibodies remains poorly understood. Although GM2 positivity has been sporadically reported in GBS, the clinical implications and associated neurological features, especially CNS involvement, have not been systematically evaluated [\u003cspan citationid=\"CR5\" class=\"CitationRef\"\u003e5\u003c/span\u003e].\u003c/p\u003e\u003cp\u003eWith the growing interest in precision medicine, machine learning has emerged as a powerful tool to uncover complex patterns within clinical and immunological datasets. In neurology, machine learning models have been applied to predict outcomes, classify disease subtypes, and identify novel biomarker associations [\u003cspan citationid=\"CR6\" class=\"CitationRef\"\u003e6\u003c/span\u003e, \u003cspan citationid=\"CR7\" class=\"CitationRef\"\u003e7\u003c/span\u003e]. In the context of GBS, machine learning approaches hold the potential to refine prognostic modeling and better interpret the relationships between clinical features and immunopathology [\u003cspan citationid=\"CR8\" class=\"CitationRef\"\u003e8\u003c/span\u003e].\u003c/p\u003e\u003cp\u003eThe present study aims to investigate the association between central nervous system signs and antiganglioside antibody profiles, particularly anti-GM2, in patients with GBS. We employed both traditional logistic regression and XGBoost-based machine learning techniques to identify clinical predictors of GM2 antibody positivity and evaluate their diagnostic performance. This work seeks to contribute to the growing understanding of CNS involvement in GBS and explore the potential of machine learning to enhance antibody prediction and guide targeted immunotherapy.\u003c/p\u003e"},{"header":"Materials and Methods","content":"\u003cdiv id=\"Sec3\" class=\"Section2\"\u003e\u003ch2\u003e1.Study Design and Participants\u003c/h2\u003e\u003cp\u003eThis retrospective study included 200 patients diagnosed with Guillain\u0026ndash;Barr\u0026eacute; syndrome (GBS) between January 2020 and December 2024 at [Institution Name]. The diagnosis of GBS was made according to the criteria proposed by the National Institute of Neurological Disorders and Stroke (NINDS), which require progressive bilateral weakness and diminished or absent deep tendon reflexes in more than one limb [\u003cspan citationid=\"CR9\" class=\"CitationRef\"\u003e9\u003c/span\u003e]. The diagnostic certainty was further supported using the Brighton criteria [\u003cspan citationid=\"CR10\" class=\"CitationRef\"\u003e10\u003c/span\u003e, \u003cspan citationid=\"CR11\" class=\"CitationRef\"\u003e11\u003c/span\u003e]. Patients with incomplete records or alternative diagnoses were excluded.\u003c/p\u003e\u003c/div\u003e\n\u003ch3\u003e2.Data Collection\u003c/h3\u003e\n\u003cp\u003eDemographic characteristics (age, sex), clinical features (e.g., Babinski sign, hyperreflexia, facial palsy), and cerebrospinal fluid (CSF) results were extracted from electronic medical records. CSF parameters included protein concentration and white blood cell count. Albuminocytologic dissociation was defined as CSF protein\u0026thinsp;\u0026gt;\u0026thinsp;0.55 g/L with \u0026lt;\u0026thinsp;10 white cells/\u0026micro;L [\u003cspan citationid=\"CR12\" class=\"CitationRef\"\u003e12\u003c/span\u003e].\u003c/p\u003e\n\u003ch3\u003e3.Antibody Testing\u003c/h3\u003e\n\u003cp\u003eSerum antiganglioside antibodies including GM1, GM2, GD1a, and GQ1b were measured using a standardized ELISA assay. Results were considered positive if the optical density exceeded the mean plus two standard deviations of healthy controls. The clinical interpretation of antibody profiles was based on previously established immunopathological correlations [\u003cspan citationid=\"CR13\" class=\"CitationRef\"\u003e13\u003c/span\u003e, \u003cspan citationid=\"CR14\" class=\"CitationRef\"\u003e14\u003c/span\u003e].\u003c/p\u003e\n\u003ch3\u003e4.Statistical Analysis\u003c/h3\u003e\n\u003cp\u003eContinuous variables were expressed as mean\u0026thinsp;\u0026plusmn;\u0026thinsp;standard deviation (SD) or median with interquartile range (IQR) depending on data distribution. Categorical variables were summarized as counts and percentages. Intergroup comparisons were performed using Student\u0026rsquo;s t-test or Mann\u0026ndash;Whitney U test for continuous variables, and chi-square or Fisher\u0026rsquo;s exact test for categorical variables, as appropriate. Logistic regression was used to identify independent predictors of GM2 antibody positivity. Odds ratios (ORs) and 95% confidence intervals (CIs) were calculated. A p-value\u0026thinsp;\u0026lt;\u0026thinsp;0.05 was considered statistically significant. All statistical analyses were conducted using R software (version 4.3.2; R Foundation for Statistical Computing, Vienna, Austria), with packages including tidyverse, pROC, xgboost, and ggplot2 [\u003cspan citationid=\"CR15\" class=\"CitationRef\"\u003e15\u003c/span\u003e].\u003c/p\u003e\n\u003ch3\u003e5.Machine Learning Modeling\u003c/h3\u003e\n\u003cp\u003eAn extreme gradient boosting (XGBoost) classifier was constructed to predict GM2 antibody positivity based on clinical features. The dataset was randomly split into training (80%) and testing (20%) sets. Model performance was evaluated using receiver operating characteristic (ROC) curves and the area under the curve (AUC). Predictor importance was visualized to identify major contributing features. The model-building process followed best practices from prior interpretable machine learning studies in GBS [\u003cspan citationid=\"CR16\" class=\"CitationRef\"\u003e16\u003c/span\u003e, \u003cspan citationid=\"CR17\" class=\"CitationRef\"\u003e17\u003c/span\u003e].\u003c/p\u003e\u003cdiv id=\"Sec8\" class=\"Section2\"\u003e\u003ch2\u003e6.Ethical Considerations\u003c/h2\u003e\u003cp\u003eThis study was approved by the Institutional Review Board of [Institution Name]. Due to the retrospective design, the requirement for informed consent was waived. All procedures complied with the Declaration of Helsinki.\u003c/p\u003e\u003cp\u003e\u003cb\u003e7.Ethics Statement and Informed Consent Statement.\u003c/b\u003e\u003c/p\u003e\u003cp\u003eThe study was a retrospective analysis and was approved by the Ethics Committee of Fujian Provincial Geriatric Hospital. The requirement for written informed consent was waived, as the study involved anonymized data and posed minimal risk to participants.\u003c/p\u003e\u003c/div\u003e\n\u003ch3\u003e8.Data Availability Statement\u003c/h3\u003e\n\u003cp\u003eThe datasets generated and analyzed during the current study are not publicly available due to ethical and privacy restrictions, but may be available from the corresponding author upon reasonable request and with permission from the institutional ethics committee.\u003c/p\u003e\u003c/li\u003e\u003c/ul\u003e\u003c/p\u003e"},{"header":"Results","content":"\u003cdiv id=\"Sec11\" class=\"Section2\"\u003e\u003ch2\u003e1. Antibody Positivity Rates\u003c/h2\u003e\u003cp\u003eAmong the 200 GBS patients, GM1 antibody positivity was most frequent (33.5%), followed by GM2 (25%), GD1a (23%), and GQ1b (21%) (Fig.\u0026nbsp;\u003cspan refid=\"Fig1\" class=\"InternalRef\"\u003e1\u003c/span\u003e, Table\u0026nbsp;\u003cspan refid=\"Tab1\" class=\"InternalRef\"\u003e1\u003c/span\u003e). These results indicate that anti-GM1 and anti-GM2 antibodies are the predominant serological markers in this cohort.\u003c/p\u003e\u003cp\u003e\u003c/p\u003e\u003cp\u003e\u003cdiv class=\"gridtable\"\u003e\u003ctable float=\"Yes\" id=\"Tab1\" border=\"1\"\u003e\u003ccaption language=\"En\"\u003e\u003cdiv class=\"CaptionNumber\"\u003eTable 1\u003c/div\u003e\u003cdiv class=\"CaptionContent\"\u003e\u003cp\u003eFrequency and percentage of antibody positivity in 200 GBS patients.\u003c/p\u003e\u003c/div\u003e\u003c/caption\u003e\u003ccolgroup cols=\"3\"\u003e\u003cdiv align=\"left\" class=\"colspec\" colname=\"c1\" colnum=\"1\"\u003e\u003c/div\u003e\u003cdiv align=\"char\" char=\".\" class=\"colspec\" colname=\"c2\" colnum=\"2\"\u003e\u003c/div\u003e\u003cdiv align=\"left\" class=\"colspec\" colname=\"c3\" colnum=\"3\"\u003e\u003c/div\u003e\u003cthead\u003e\u003ctr\u003e\u003cth align=\"left\" colname=\"c1\"\u003e\u003cp\u003eAntibody\u003c/p\u003e\u003c/th\u003e\u003cth align=\"left\" colname=\"c2\"\u003e\u003cp\u003eFrequency(%)\u003c/p\u003e\u003c/th\u003e\u003cth align=\"left\" colname=\"c3\"\u003e\u003cp\u003ePercentage(%)\u003c/p\u003e\u003c/th\u003e\u003c/tr\u003e\u003c/thead\u003e\u003ctbody\u003e\u003ctr\u003e\u003ctd align=\"left\" colname=\"c1\"\u003e\u003cp\u003eGD1a_Pos\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"char\" char=\".\" colname=\"c2\"\u003e\u003cp\u003e46\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c3\"\u003e\u003cp\u003e23\u003c/p\u003e\u003c/td\u003e\u003c/tr\u003e\u003ctr\u003e\u003ctd align=\"left\" colname=\"c1\"\u003e\u003cp\u003eGM1_Pos\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"char\" char=\".\" colname=\"c2\"\u003e\u003cp\u003e67\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c3\"\u003e\u003cp\u003e33.5\u003c/p\u003e\u003c/td\u003e\u003c/tr\u003e\u003ctr\u003e\u003ctd align=\"left\" colname=\"c1\"\u003e\u003cp\u003eGM2_Pos\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"char\" char=\".\" colname=\"c2\"\u003e\u003cp\u003e50\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c3\"\u003e\u003cp\u003e25\u003c/p\u003e\u003c/td\u003e\u003c/tr\u003e\u003ctr\u003e\u003ctd align=\"left\" colname=\"c1\"\u003e\u003cp\u003eGQ1b_Pos\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"char\" char=\".\" colname=\"c2\"\u003e\u003cp\u003e42\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c3\"\u003e\u003cp\u003e21\u003c/p\u003e\u003c/td\u003e\u003c/tr\u003e\u003c/tbody\u003e\u003c/colgroup\u003e\u003c/table\u003e\u003c/div\u003e\u003c/p\u003e\u003c/div\u003e\u003cdiv id=\"Sec12\" class=\"Section2\"\u003e\u003ch2\u003e2. Association Between Central Nervous System Signs and GM2 Positivity\u003c/h2\u003e\u003cp\u003ePatients exhibiting central nervous system signs were more likely to test positive for GM2 antibodies. Specifically, 35 of the 50 GM2-positive cases (70%) presented with central signs, compared to 40 of 150 GM2-negative patients (26.7%) (Fig.\u0026nbsp;\u003cspan refid=\"Fig2\" class=\"InternalRef\"\u003e2\u003c/span\u003e, Table\u0026nbsp;\u003cspan refid=\"Tab2\" class=\"InternalRef\"\u003e2\u003c/span\u003e). This suggests a potential link between central signs and GM2 antibody presence.\u003c/p\u003e\u003cp\u003e\u003c/p\u003e\u003cp\u003e\u003cdiv class=\"gridtable\"\u003e\u003ctable float=\"Yes\" id=\"Tab2\" border=\"1\"\u003e\u003ccaption language=\"En\"\u003e\u003cdiv class=\"CaptionNumber\"\u003eTable 2\u003c/div\u003e\u003cdiv class=\"CaptionContent\"\u003e\u003cp\u003eDistribution of central signs by GM2 antibody status.\u003c/p\u003e\u003c/div\u003e\u003c/caption\u003e\u003ccolgroup cols=\"3\"\u003e\u003cdiv align=\"left\" class=\"colspec\" colname=\"c1\" colnum=\"1\"\u003e\u003c/div\u003e\u003cdiv align=\"left\" class=\"colspec\" colname=\"c2\" colnum=\"2\"\u003e\u003c/div\u003e\u003cdiv align=\"char\" char=\".\" class=\"colspec\" colname=\"c3\" colnum=\"3\"\u003e\u003c/div\u003e\u003cthead\u003e\u003ctr\u003e\u003cth align=\"left\" colname=\"c1\"\u003e\u003cp\u003eCentral_Signs\u003c/p\u003e\u003c/th\u003e\u003cth align=\"left\" colname=\"c2\"\u003e\u003cp\u003eGM2_Pos\u003c/p\u003e\u003c/th\u003e\u003cth align=\"left\" colname=\"c3\"\u003e\u003cp\u003eCount\u003c/p\u003e\u003c/th\u003e\u003c/tr\u003e\u003c/thead\u003e\u003ctbody\u003e\u003ctr\u003e\u003ctd align=\"left\" colname=\"c1\"\u003e\u003cp\u003eNo\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c2\"\u003e\u003cp\u003eNo\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"char\" char=\".\" colname=\"c3\"\u003e\u003cp\u003e110\u003c/p\u003e\u003c/td\u003e\u003c/tr\u003e\u003ctr\u003e\u003ctd align=\"left\" colname=\"c1\"\u003e\u003cp\u003eYes\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c2\"\u003e\u003cp\u003eNo\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"char\" char=\".\" colname=\"c3\"\u003e\u003cp\u003e40\u003c/p\u003e\u003c/td\u003e\u003c/tr\u003e\u003ctr\u003e\u003ctd align=\"left\" colname=\"c1\"\u003e\u003cp\u003eNo\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c2\"\u003e\u003cp\u003eYes\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"char\" char=\".\" colname=\"c3\"\u003e\u003cp\u003e15\u003c/p\u003e\u003c/td\u003e\u003c/tr\u003e\u003ctr\u003e\u003ctd align=\"left\" colname=\"c1\"\u003e\u003cp\u003eYes\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c2\"\u003e\u003cp\u003eYes\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"char\" char=\".\" colname=\"c3\"\u003e\u003cp\u003e35\u003c/p\u003e\u003c/td\u003e\u003c/tr\u003e\u003c/tbody\u003e\u003c/colgroup\u003e\u003c/table\u003e\u003c/div\u003e\u003c/p\u003e\u003c/div\u003e\u003cdiv id=\"Sec13\" class=\"Section2\"\u003e\u003ch2\u003e3. Differences in Clinical Variables by GM2 Antibody Status\u003c/h2\u003e\u003cp\u003eCompared to GM2-negative patients, those with GM2 positivity showed higher CSF protein levels (Fig.\u0026nbsp;\u003cspan refid=\"Fig3\" class=\"InternalRef\"\u003e3\u003c/span\u003e) and a trend toward younger age. However, no significant difference was observed in CSF white cell counts between the two groups (Fig.\u0026nbsp;\u003cspan refid=\"Fig4\" class=\"InternalRef\"\u003e4\u003c/span\u003e).\u003c/p\u003e\u003cp\u003e\u003c/p\u003e\u003cp\u003e\u003c/p\u003e\u003c/div\u003e\u003cdiv id=\"Sec14\" class=\"Section2\"\u003e\u003ch2\u003e4. Neurological Sign Patterns\u003c/h2\u003e\u003cp\u003eA heatmap was generated to visualize the distribution of key neurological signs. Babinski sign and hyperreflexia were more frequently observed in the GM2-positive group, whereas facial palsy and hyporeflexia were less discriminative (Fig.\u0026nbsp;\u003cspan refid=\"Fig5\" class=\"InternalRef\"\u003e5\u003c/span\u003e).\u003c/p\u003e\u003cp\u003e\u003c/p\u003e\u003c/div\u003e\u003cdiv id=\"Sec15\" class=\"Section2\"\u003e\u003ch2\u003e5. Correlation Between Continuous Variables\u003c/h2\u003e\u003cp\u003eCorrelation analysis revealed no strong associations among age, CSF protein, and CSF white blood cell count (Fig.\u0026nbsp;\u003cspan refid=\"Fig6\" class=\"InternalRef\"\u003e6\u003c/span\u003e). The modest correlations suggest these variables are largely independent predictors.\u003c/p\u003e\u003cp\u003e\u003c/p\u003e\u003c/div\u003e\u003cdiv id=\"Sec16\" class=\"Section2\"\u003e\u003ch2\u003e6. Logistic Regression Analysis for Predicting GM2 Positivity\u003c/h2\u003e\u003cp\u003eMultivariate logistic regression identified central nervous system signs (OR\u0026thinsp;=\u0026thinsp;7.92, 95% CI: 3.77\u0026ndash;17.76) and Babinski sign (OR\u0026thinsp;=\u0026thinsp;3.16, 95% CI: 1.39\u0026ndash;7.38) as independent predictors of GM2 antibody positivity. Age and CSF protein level did not reach statistical significance (Fig.\u0026nbsp;\u003cspan refid=\"Fig7\" class=\"InternalRef\"\u003e7\u003c/span\u003e, Table\u0026nbsp;\u003cspan refid=\"Tab3\" class=\"InternalRef\"\u003e3\u003c/span\u003e).\u003c/p\u003e\u003cp\u003e\u003c/p\u003e\u003cp\u003e\u003cdiv class=\"gridtable\"\u003e\u003ctable float=\"Yes\" id=\"Tab3\" border=\"1\"\u003e\u003ccaption language=\"En\"\u003e\u003cdiv class=\"CaptionNumber\"\u003eTable 3\u003c/div\u003e\u003cdiv class=\"CaptionContent\"\u003e\u003cp\u003eLogistic regression analysis of clinical predictors for GM2 antibody positivity.\u003c/p\u003e\u003c/div\u003e\u003c/caption\u003e\u003ccolgroup cols=\"4\"\u003e\u003cdiv align=\"left\" class=\"colspec\" colname=\"c1\" colnum=\"1\"\u003e\u003c/div\u003e\u003cdiv align=\"char\" char=\".\" class=\"colspec\" colname=\"c2\" colnum=\"2\"\u003e\u003c/div\u003e\u003cdiv align=\"char\" char=\".\" class=\"colspec\" colname=\"c3\" colnum=\"3\"\u003e\u003c/div\u003e\u003cdiv align=\"char\" char=\".\" class=\"colspec\" colname=\"c4\" colnum=\"4\"\u003e\u003c/div\u003e\u003cthead\u003e\u003ctr\u003e\u003cth align=\"left\" colname=\"c1\"\u003e\u003cp\u003eVariable\u003c/p\u003e\u003c/th\u003e\u003cth align=\"left\" colname=\"c2\"\u003e\u003cp\u003eOdds_Ratio\u003c/p\u003e\u003c/th\u003e\u003cth align=\"left\" colname=\"c3\"\u003e\u003cp\u003eX2.5..\u003c/p\u003e\u003c/th\u003e\u003cth align=\"left\" colname=\"c4\"\u003e\u003cp\u003eX97.5..\u003c/p\u003e\u003c/th\u003e\u003c/tr\u003e\u003c/thead\u003e\u003ctbody\u003e\u003ctr\u003e\u003ctd align=\"left\" colname=\"c1\"\u003e\u003cp\u003e(Intercept)\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"char\" char=\".\" colname=\"c2\"\u003e\u003cp\u003e0.032813334\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"char\" char=\".\" colname=\"c3\"\u003e\u003cp\u003e0.002097196\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"char\" char=\".\" colname=\"c4\"\u003e\u003cp\u003e0.440287899\u003c/p\u003e\u003c/td\u003e\u003c/tr\u003e\u003ctr\u003e\u003ctd align=\"left\" colname=\"c1\"\u003e\u003cp\u003eCentral_SignsYes\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"char\" char=\".\" colname=\"c2\"\u003e\u003cp\u003e7.9190422\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"char\" char=\".\" colname=\"c3\"\u003e\u003cp\u003e3.770400489\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"char\" char=\".\" colname=\"c4\"\u003e\u003cp\u003e17.75595832\u003c/p\u003e\u003c/td\u003e\u003c/tr\u003e\u003ctr\u003e\u003ctd align=\"left\" colname=\"c1\"\u003e\u003cp\u003eBabinski_SignYes\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"char\" char=\".\" colname=\"c2\"\u003e\u003cp\u003e3.155231613\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"char\" char=\".\" colname=\"c3\"\u003e\u003cp\u003e1.393877438\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"char\" char=\".\" colname=\"c4\"\u003e\u003cp\u003e7.381464718\u003c/p\u003e\u003c/td\u003e\u003c/tr\u003e\u003ctr\u003e\u003ctd align=\"left\" colname=\"c1\"\u003e\u003cp\u003eAge\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"char\" char=\".\" colname=\"c2\"\u003e\u003cp\u003e0.989564128\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"char\" char=\".\" colname=\"c3\"\u003e\u003cp\u003e0.971874474\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"char\" char=\".\" colname=\"c4\"\u003e\u003cp\u003e1.00724218\u003c/p\u003e\u003c/td\u003e\u003c/tr\u003e\u003ctr\u003e\u003ctd align=\"left\" colname=\"c1\"\u003e\u003cp\u003e`CSF_Protein (g/L)`\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"char\" char=\".\" colname=\"c2\"\u003e\u003cp\u003e11.78591297\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"char\" char=\".\" colname=\"c3\"\u003e\u003cp\u003e0.292191333\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"char\" char=\".\" colname=\"c4\"\u003e\u003cp\u003e546.3601361\u003c/p\u003e\u003c/td\u003e\u003c/tr\u003e\u003c/tbody\u003e\u003c/colgroup\u003e\u003c/table\u003e\u003c/div\u003e\u003c/p\u003e\u003c/div\u003e\u003cdiv id=\"Sec17\" class=\"Section2\"\u003e\u003ch2\u003e7. Predictive Performance of Logistic vs. Machine Learning Models\u003c/h2\u003e\u003cp\u003eBoth the logistic regression and XGBoost models demonstrated similar discriminative ability for predicting GM2 positivity, with AUC values of 0.741 and 0.735, respectively (Fig.\u0026nbsp;\u003cspan refid=\"Fig8\" class=\"InternalRef\"\u003e8\u003c/span\u003e). AUC values for both models were summarized for direct visual comparison (Fig.\u0026nbsp;\u003cspan refid=\"Fig9\" class=\"InternalRef\"\u003e9\u003c/span\u003e). This suggests that traditional statistical models remain competitive with machine learning approaches in this context.\u003c/p\u003e\u003c/li\u003e\u003c/span\u003e\u003c/ol\u003e\u003c/p\u003e\u003cp\u003e\u003c/p\u003e\u003cp\u003e\u003c/p\u003e\u003c/div\u003e\u003cdiv id=\"Sec18\" class=\"Section2\"\u003e\u003ch2\u003e8. Feature Importance in XGBoost Model\u003c/h2\u003e\u003cp\u003eThe top contributing features in the XGBoost model included central nervous system signs, Babinski sign, hyperreflexia, and CSF protein concentration (Fig.\u0026nbsp;\u003cspan refid=\"Fig10\" class=\"InternalRef\"\u003e10\u003c/span\u003e). These findings support the clinical relevance of these features in antibody prediction.\u003c/p\u003e\u003cp\u003e\u003c/p\u003e\u003c/div\u003e"},{"header":"Discussion","content":"\u003cp\u003eIn this study, we demonstrated that central nervous system (CNS) signs, particularly Babinski sign and generalized hyperreflexia, were significantly associated with GM2 antibody positivity in patients with Guillain\u0026ndash;Barr\u0026eacute; syndrome (GBS). This finding adds new insight to the traditional view of GBS as a peripheral neuropathy, supporting the concept that certain immunological subtypes may also involve central motor pathways [\u003cspan citationid=\"CR18\" class=\"CitationRef\"\u003e18\u003c/span\u003e, \u003cspan citationid=\"CR19\" class=\"CitationRef\"\u003e19\u003c/span\u003e].\u003c/p\u003e\u003cp\u003eThe GM2 antibody positivity rate in our cohort reached 25%, which is higher than previously reported estimates ranging from 5\u0026ndash;15% in unselected GBS populations [\u003cspan citationid=\"CR20\" class=\"CitationRef\"\u003e20\u003c/span\u003e, \u003cspan citationid=\"CR21\" class=\"CitationRef\"\u003e21\u003c/span\u003e]. One possible explanation is the relatively broad inclusion of GBS variants with CNS features and improved detection sensitivity of current ELISA-based assays. These findings support earlier suggestions that GM2 may play a broader pathogenic role beyond cranial nerve involvement [\u003cspan citationid=\"CR22\" class=\"CitationRef\"\u003e22\u003c/span\u003e].\u003c/p\u003e\u003cp\u003eImportantly, CNS signs and Babinski sign emerged as strong independent predictors of GM2 positivity in our multivariate logistic regression model, with odds ratios of 7.92 and 3.16, respectively. These findings emphasize that careful neurological examination may provide meaningful clues for antibody profiling, particularly when laboratory resources are limited. In contrast, traditional biomarkers such as CSF protein concentration and age, although showing trends in univariate comparisons, did not retain significance in multivariate analysis, suggesting that GM2 positivity may reflect a more specific pathophysiological mechanism rather than general inflammation [\u003cspan citationid=\"CR23\" class=\"CitationRef\"\u003e23\u003c/span\u003e, \u003cspan citationid=\"CR24\" class=\"CitationRef\"\u003e24\u003c/span\u003e].\u003c/p\u003e\u003cp\u003ePrevious studies have demonstrated that anti-GM2 antibodies can bind to neural gangliosides distributed in both peripheral and central axons [\u003cspan citationid=\"CR25\" class=\"CitationRef\"\u003e25\u003c/span\u003e], potentially contributing to upper motor neuron signs observed in our patients. This aligns with known mechanisms in anti-GQ1b syndromes, where similar antibody-mediated disruption of nodal or paranodal regions can result in overlapping features of central and peripheral involvement [\u003cspan citationid=\"CR26\" class=\"CitationRef\"\u003e26\u003c/span\u003e, \u003cspan citationid=\"CR31\" class=\"CitationRef\"\u003e31\u003c/span\u003e].\u003c/p\u003e\u003cp\u003eBoth logistic regression and the XGBoost machine learning model achieved comparable performance in predicting GM2 positivity, with AUC values of 0.741 and 0.735, respectively. Although machine learning algorithms are often promoted for their superior ability to detect nonlinear interactions, our results suggest that when appropriate variables are chosen, classical statistical models can yield similarly robust discrimination while retaining greater interpretability [\u003cspan citationid=\"CR28\" class=\"CitationRef\"\u003e28\u003c/span\u003e, \u003cspan citationid=\"CR29\" class=\"CitationRef\"\u003e29\u003c/span\u003e]. The tree-based nature of XGBoost also allows for handling missing data and non-linear interactions, which can be advantageous in complex clinical datasets [\u003cspan citationid=\"CR29\" class=\"CitationRef\"\u003e29\u003c/span\u003e].\u003c/p\u003e\u003cp\u003eThe XGBoost model additionally revealed feature importance rankings that confirmed the dominant role of CNS signs and Babinski sign in GM2 prediction. These machine-driven rankings corroborate the regression-based findings, supporting the notion that even data-driven models echo fundamental clinical observations.\u003c/p\u003e\u003cp\u003eThe identification of these clinical predictors may have practical implications. In many settings, serological testing for antiganglioside antibodies is either unavailable or delayed. Recognizing CNS signs at the bedside could help prioritize testing for anti-GM2 antibodies and guide earlier initiation of immunotherapy. Moreover, incorporating these variables into algorithmic models may assist emergency and general neurologists in early decision-making, especially for atypical GBS presentations. Future work may incorporate neural network models or deep learning frameworks to explore subtle patterns not captured by current models [\u003cspan citationid=\"CR30\" class=\"CitationRef\"\u003e30\u003c/span\u003e].\u003c/p\u003e\u003cp\u003eNonetheless, several limitations should be acknowledged. First, the retrospective nature of our study may introduce selection bias and limit the assessment of causality. Second, we did not evaluate serial antibody titers or long-term outcomes, precluding conclusions regarding prognosis. Third, although the sample size is relatively large for a single-center cohort, our findings require validation in independent and multicenter settings. Furthermore, heterogeneity in ELISA assay platforms across centers may affect antibody detection rates and their reported clinical correlations [\u003cspan citationid=\"CR32\" class=\"CitationRef\"\u003e32\u003c/span\u003e].\u003c/p\u003e"},{"header":"Conclusion","content":"\u003cp\u003eThis study provides evidence that central nervous system signs, particularly Babinski sign and generalized hyperreflexia, are independently associated with GM2 antibody positivity in patients with Guillain\u0026ndash;Barr\u0026eacute; syndrome. These findings suggest that clinical signs traditionally considered outside the scope of peripheral neuropathy may, in fact, serve as relevant indicators of specific immune subtypes. Both logistic regression and machine learning approaches demonstrated reliable predictive performance, with XGBoost offering additional insight into feature importance.\u003c/p\u003e\u003cp\u003eFrom a clinical standpoint, bedside identification of CNS signs may help prioritize antibody screening, particularly when testing resources are limited. Our results also support the potential integration of interpretable machine learning models into clinical workflows for antibody prediction in GBS. Further prospective, multicenter studies are needed to validate these findings and explore whether GM2-positive GBS represents a distinct clinico-immunological subtype.\u003c/p\u003e"},{"header":"Declarations","content":"\u003ch2\u003eAuthor Contribution\u003c/h2\u003e\u003cp\u003eAuthor Contributions: J.L. was solely responsible for the conception and design of the study, data analysis, figure preparation, interpretation of results, and drafting and revision of the manuscript.\u003c/p\u003e\u003ch2\u003eData Availability\u003c/h2\u003e\u003cp\u003eThe datasets generated and/or analyzed during the current study are not publicly available due to ethical and privacy restrictions, but may be made available by the corresponding author on reasonable request and with approval from the Ethics Committee of Fujian Provincial Geriatric Hospital.\u003c/p\u003e"},{"header":"References","content":"\u003col\u003e\n\u003cli\u003eWakerley, B. R. \u0026amp; Yuki, N. Mimics and chameleons in Guillain–Barré and Miller Fisher syndromes. \u003cem\u003ePract. Neurol.\u003c/em\u003e \u003cb\u003e15\u003c/b\u003e (2), 90–99. https://doi.org/10.1136/practneurol-2014-001101(2015).\u003c/li\u003e\n\u003cli\u003eYuki, N. \u0026amp; Hartung, H. P. Guillain–Barré syndrome. \u003cem\u003eN Engl. J. Med.\u003c/em\u003e \u003cb\u003e366\u003c/b\u003e (24), 2294–2304. https://doi.org/10.1056/NEJMra1114525(2012).\u003c/li\u003e\n\u003cli\u003eWillison, H. J., Jacobs, B. C. \u0026amp; van Doorn, P. A. Guillain–Barré syndrome. \u003cem\u003eLancet\u003c/em\u003e \u003cb\u003e388\u003c/b\u003e (10045), 717–727. https://doi.org/10.1016/S0140-6736(16)00339-1(2016).\u003c/li\u003e\n\u003cli\u003eKaida, K., Ariga, T. \u0026amp; Yu, R. K. Antiganglioside antibodies and their pathophysiological effects on Guillain–Barré syndrome and related disorders—a review. \u003cem\u003eGlycobiology\u003c/em\u003e \u003cb\u003e19\u003c/b\u003e (7), 676–692. https://doi.org/10.1093/glycob/cwp048(2009).\u003c/li\u003e\n\u003cli\u003eKusunoki, S. \u0026amp; Kaida, K. Antibodies against ganglioside complexes in Guillain–Barré syndrome and related disorders. \u003cem\u003eJ. Neurochem\u003c/em\u003e. \u003cb\u003e116\u003c/b\u003e (5), 828–832. https://doi.org/10.1111/j.1471-4159.2010.07161.x(2011).\u003c/li\u003e\n\u003cli\u003eZhang, Y. et al. Interpretable machine learning model for predicting the prognosis of Guillain–Barré syndrome patients. \u003cem\u003eJ. Inflamm. Res.\u003c/em\u003e \u003cb\u003e16\u003c/b\u003e, 471–482. https://doi.org/10.2147/JIR.S391028(2023).\u003c/li\u003e\n\u003cli\u003eSantoro, M., Manganelli, F. \u0026amp; Iodice, R. Artificial intelligence and machine learning in peripheral neuropathies: current applications and future perspectives. \u003cem\u003eFront. Neurol.\u003c/em\u003e \u003cb\u003e13\u003c/b\u003e, 906984. https://doi.org/10.3389/fneur.2022.906984(2022).\u003c/li\u003e\n\u003cli\u003eZhang, J. et al. Identification of clinical subtypes and immune-related biomarkers in GBS using machine learning. \u003cem\u003eFront. Immunol.\u003c/em\u003e \u003cb\u003e14\u003c/b\u003e, 1143761. https://doi.org/10.3389/fimmu.2023.1143761(2023).\u003c/li\u003e\n\u003cli\u003eAsbury, A. K. \u0026amp; Cornblath, D. R. Assessment of current diagnostic criteria for Guillain–Barré syndrome. \u003cem\u003eAnn. Neurol.\u003c/em\u003e \u003cb\u003e27\u003c/b\u003e (Suppl), S21–S24. https://doi.org/10.1002/ana.410270707(1990).\u003c/li\u003e\n\u003cli\u003ean den .Fokke, C. et al. Diagnosis of Guillain–Barré syndrome and validation of Brighton criteria. \u003cem\u003eBrain\u003c/em\u003e \u003cb\u003e137\u003c/b\u003e (Pt 1), 33–43. https://doi.org/10.1093/brain/awt285(2014).\u003c/li\u003e\n\u003cli\u003eWakerley, B. R., Uncini, A. \u0026amp; Yuki, N. Guillain–Barré and Miller Fisher syndromes—new diagnostic classification. \u003cem\u003eNat. Rev. Neurol.\u003c/em\u003e \u003cb\u003e10\u003c/b\u003e (9), 537–544. https://doi.org/10.1038/nrneurol.2014.138(2014).\u003c/li\u003e\n\u003cli\u003eHughes, R. A. et al. Supportive care for patients with Guillain–Barré syndrome. \u003cem\u003eArch. Neurol.\u003c/em\u003e \u003cb\u003e62\u003c/b\u003e (8), 1194–1198. https://doi.org/10.1001/archneur.62.8.1194(2005).\u003c/li\u003e\n\u003cli\u003eKaida, K., Ariga, T. \u0026amp; Yu, R. K. Antiganglioside antibodies and their pathophysiological effects on Guillain–Barré syndrome and related disorders—a review. \u003cem\u003eGlycobiology\u003c/em\u003e \u003cb\u003e19\u003c/b\u003e (7), 676–692. https://doi.org/10.1093/glycob/cwp048(2009).\u003c/li\u003e\n\u003cli\u003eWillison, H. J., Jacobs, B. C. \u0026amp; van Doorn, P. A. Guillain–Barré syndrome. \u003cem\u003eLancet\u003c/em\u003e \u003cb\u003e388\u003c/b\u003e (10045), 717–727. https://doi.org/10.1016/S0140-6736(16)00339-1(2016).\u003c/li\u003e\n\u003cli\u003eore Team, R. R: A language and environment for statistical computing. R Foundation for Statistical Computing, Vienna, Austria. https://www.R-project.org/\u003c/li\u003e\n\u003cli\u003eZhang, Y. et al. Interpretable machine learning model for predicting the prognosis of Guillain–Barré syndrome patients. \u003cem\u003eJ. Inflamm. Res.\u003c/em\u003e \u003cb\u003e16\u003c/b\u003e, 471–482. https://doi.org/10.2147/JIR.S391028(2023).\u003c/li\u003e\n\u003cli\u003eChen, J. J., Hsu, C. W. \u0026amp; Shen, C. J. Biomarkers and machine learning in Guillain–Barré syndrome. \u003cem\u003eClin. Chim. Acta\u003c/em\u003e. \u003cb\u003e519\u003c/b\u003e, 106–111. https://doi.org/10.1016/j.cca.2021.04.023(2021).\u003c/li\u003e\n\u003cli\u003eKuwabara, S. et al. IgG anti-GM1 antibody is associated with reversible conduction failure and axonal degeneration in Guillain-Barré syndrome. \u003cem\u003eAnn. Neurol.\u003c/em\u003e \u003cb\u003e44\u003c/b\u003e (2), 202–208. https://doi.org/10.1002/ana.410440211(1998).\u003c/li\u003e\n\u003cli\u003eWakerley, B. R., Uncini, A. \u0026amp; Yuki, N. Guillain–Barré and Miller Fisher syndromes—new diagnostic classification. \u003cem\u003eNat. Rev. Neurol.\u003c/em\u003e \u003cb\u003e10\u003c/b\u003e (9), 537–544. https://doi.org/10.1038/nrneurol.2014.138(2014).\u003c/li\u003e\n\u003cli\u003eKusunoki, S., Chiba, A. \u0026amp; Kanazawa, I. Anti-GM2 antibody in Guillain–Barré syndrome. \u003cem\u003eMuscle Nerve\u003c/em\u003e. \u003cb\u003e20\u003c/b\u003e (11), 1372–1374. https://doi.org/10.1002/(SICI)1097-4598(199711)20:11\u0026lt;1372::AID-MUS10\u0026gt;3.0.CO;2-Z(1997).\u003c/li\u003e\n\u003cli\u003eKaida, K., Ariga, T. \u0026amp; Yu, R. K. Antiganglioside antibodies and their pathophysiological effects on Guillain–Barré syndrome and related disorders—a review. \u003cem\u003eGlycobiology\u003c/em\u003e \u003cb\u003e19\u003c/b\u003e (7), 676–692. https://doi.org/10.1093/glycob/cwp048(2009).\u003c/li\u003e\n\u003cli\u003e den Berg, B. et al. Clinical outcome and antibody profiles in anti-GM2 positive Guillain–Barré syndrome. \u003cem\u003eJ. Neurol. Neurosurg. Psychiatry\u003c/em\u003e. \u003cb\u003e85\u003c/b\u003e (7), 843–846. https://doi.org/10.1136/jnnp-2013-307054(2014).\u003c/li\u003e\n\u003cli\u003eHughes, R. A. et al. Supportive care for patients with Guillain–Barré syndrome. \u003cem\u003eArch. Neurol.\u003c/em\u003e \u003cb\u003e62\u003c/b\u003e (8), 1194–1198. https://doi.org/10.1001/archneur.62.8.1194(2005).\u003c/li\u003e\n\u003cli\u003ean den .Fokke, C. et al. Diagnosis of Guillain–Barré syndrome and validation of Brighton criteria. \u003cem\u003eBrain\u003c/em\u003e \u003cb\u003e137\u003c/b\u003e (Pt 1), 33–43. https://doi.org/10.1093/brain/awt285(2014).\u003c/li\u003e\n\u003cli\u003eWillison, H. J., Jacobs, B. C. \u0026amp; van Doorn, P. A. Guillain–Barré syndrome. \u003cem\u003eLancet\u003c/em\u003e \u003cb\u003e388\u003c/b\u003e (10045), 717–727. https://doi.org/10.1016/S0140-6736(16)00339-1(2016).\u003c/li\u003e\n\u003cli\u003eYuki, N. et al. A bacterium lipopolysaccharide that elicits Guillain–Barré syndrome has a GM1 ganglioside-like structure. \u003cem\u003eJ. Exp. Med.\u003c/em\u003e \u003cb\u003e178\u003c/b\u003e (5), 1771–1775. https://doi.org/10.1084/jem.178.5.1771(1993).\u003c/li\u003e\n\u003cli\u003eZhang, Y. et al. Interpretable machine learning model for predicting the prognosis of Guillain–Barré syndrome patients. \u003cem\u003eJ. Inflamm. Res.\u003c/em\u003e \u003cb\u003e16\u003c/b\u003e, 471–482. https://doi.org/10.2147/JIR.S391028(2023).\u003c/li\u003e\n\u003cli\u003eChristodoulou, E. et al. A systematic review shows no performance benefit of machine learning over logistic regression for clinical prediction models. \u003cem\u003eJ. Clin. Epidemiol.\u003c/em\u003e \u003cb\u003e110\u003c/b\u003e, 12–22. https://doi.org/10.1016/j.jclinepi.2019.02.004(2019).\u003c/li\u003e\n\u003cli\u003eChen, J. J., Hsu, C. W. \u0026amp; Shen, C. J. Biomarkers and machine learning in Guillain–Barré syndrome. \u003cem\u003eClin. Chim. Acta\u003c/em\u003e. \u003cb\u003e519\u003c/b\u003e, 106–111. https://doi.org/10.1016/j.cca.2021.04.023(2021).\u003c/li\u003e\n\u003cli\u003eRajkomar, A., Dean, J. \u0026amp; Kohane, I. Machine learning in medicine. \u003cem\u003eN Engl. J. Med.\u003c/em\u003e \u003cb\u003e380\u003c/b\u003e (14), 1347–1358. https://doi.org/10.1056/NEJMra1814259(2019).\u003c/li\u003e\n\u003cli\u003eWillison, H. J. The immunobiology of Guillain–Barré syndromes. \u003cem\u003eJ. Peripher Nerv. Syst.\u003c/em\u003e \u003cb\u003e10\u003c/b\u003e (2), 94–112. https://doi.org/10.1111/j.1085-9489.2005.10304.x(2005).\u003c/li\u003e\n\u003cli\u003eWijdicks, E. F. M. \u0026amp; Klein, C. J. Guillain–Barré syndrome. Mayo Clin Proc. ;92(3):467–479. (2017). https://doi.org/10.1016/j.mayocp.2016.11.007\u003c/li\u003e\n\u003c/ol\u003e"}],"fulltextSource":"","fullText":"","funders":[],"hasAdminPriorityOnWorkflow":false,"hasManuscriptDocX":true,"hasOptedInToPreprint":true,"hasPassedJournalQc":"","hasAnyPriority":false,"hideJournal":true,"highlight":"","institution":"","isAcceptedByJournal":false,"isAuthorSuppliedPdf":false,"isDeskRejected":"","isHiddenFromSearch":false,"isInQc":false,"isInWorkflow":false,"isPdf":false,"isPdfUpToDate":true,"isWithdrawnOrRetracted":false,"journal":{"display":true,"email":"[email protected]","identity":"researchsquare","isNatureJournal":false,"hasQc":true,"allowDirectSubmit":true,"externalIdentity":"","sideBox":"","snPcode":"","submissionUrl":"/submission","title":"Research Square","twitterHandle":"researchsquare","acdcEnabled":true,"dfaEnabled":false,"editorialSystem":"","reportingPortfolio":"","inReviewEnabled":false,"inReviewRevisionsEnabled":true},"keywords":"Guillain–Barré syndrome, central nervous system signs, GM2 antibody, machine learning, logistic regression, XGBoost","lastPublishedDoi":"10.21203/rs.3.rs-6546968/v1","lastPublishedDoiUrl":"https://doi.org/10.21203/rs.3.rs-6546968/v1","license":{"name":"CC BY 4.0","url":"https://creativecommons.org/licenses/by/4.0/"},"manuscriptAbstract":"\u003ch2\u003eObjective\u003c/h2\u003e\u003cp\u003eTo investigate the association between central nervous system (CNS) signs and antiganglioside antibody profiles in patients with Guillain\u0026ndash;Barr\u0026eacute; syndrome (GBS), and to evaluate the predictive value of clinical features for GM2 antibody positivity using logistic regression and machine learning models.\u003c/p\u003e\u003ch2\u003eMethods\u003c/h2\u003e\u003cp\u003eA total of 200 patients with GBS were retrospectively analyzed. Clinical data including age, neurological signs, cerebrospinal fluid parameters, and antiganglioside antibody results were collected. Logistic regression was used to assess independent predictors of GM2 antibody positivity. An XGBoost model was constructed to evaluate predictive performance. Model discrimination was assessed by receiver operating characteristic (ROC) curves and area under the curve (AUC) metrics. Feature importance was analyzed to interpret model behavior.\u003c/p\u003e\u003ch2\u003eResults\u003c/h2\u003e\u003cp\u003eAmong the 200 patients, the positivity rates for GM1, GM2, GD1a, and GQ1b antibodies were 33.5%, 25%, 23%, and 21%, respectively. Central signs were observed in 35 of the 50 GM2-positive patients. Logistic regression identified central signs (OR\u0026thinsp;=\u0026thinsp;7.92, 95% CI: 3.77\u0026ndash;17.76) and Babinski sign (OR\u0026thinsp;=\u0026thinsp;3.16, 95% CI: 1.39\u0026ndash;7.38) as independent predictors of GM2 positivity. The XGBoost model achieved comparable discrimination (AUC\u0026thinsp;=\u0026thinsp;0.735) to the logistic regression model (AUC\u0026thinsp;=\u0026thinsp;0.741). Feature importance analysis revealed central signs and Babinski sign as dominant contributors.\u003c/p\u003e\u003ch2\u003eConclusion\u003c/h2\u003e\u003cp\u003eCNS signs, particularly the presence of Babinski sign and generalized hyperreflexia, are significantly associated with GM2 antibody positivity in GBS. These features may serve as practical clinical indicators to prompt antibody testing and guide immunotherapy decisions.\u003c/p\u003e","manuscriptTitle":"Central Signs Predict Antiganglioside Antibodies in Guillain–Barré Syndrome: A Machine Learning-Based Analysis","msid":"","msnumber":"","nonDraftVersions":[{"code":1,"date":"2025-08-20 06:14:42","doi":"10.21203/rs.3.rs-6546968/v1","editorialEvents":[{"type":"communityComments","content":0}],"status":"published","journal":{"display":true,"email":"[email protected]","identity":"researchsquare","isNatureJournal":false,"hasQc":true,"allowDirectSubmit":true,"externalIdentity":"","sideBox":"","snPcode":"","submissionUrl":"/submission","title":"Research Square","twitterHandle":"researchsquare","acdcEnabled":true,"dfaEnabled":false,"editorialSystem":"","reportingPortfolio":"","inReviewEnabled":false,"inReviewRevisionsEnabled":true}}],"origin":"","ownerIdentity":"a7b02483-5d15-48d7-943f-707f0e51ae05","owner":[],"postedDate":"August 20th, 2025","published":true,"recentEditorialEvents":[],"rejectedJournal":[],"revision":"","amendment":"","status":"posted","subjectAreas":[{"id":53250598,"name":"Health sciences/Neurology"},{"id":53250599,"name":"Health sciences/Signs and symptoms"}],"tags":[],"updatedAt":"2026-01-09T08:55:06+00:00","versionOfRecord":[],"versionCreatedAt":"2025-08-20 06:14:42","video":"","vorDoi":"","vorDoiUrl":"","workflowStages":[]},"version":"v1","identity":"rs-6546968","journalConfig":"researchsquare"},"__N_SSP":true},"page":"/article/[identity]/[[...version]]","query":{"redirect":"/article/rs-6546968","identity":"rs-6546968","version":["v1"]},"buildId":"XKTyCvWXoU3ODBz1xrDgd","isFallback":false,"isExperimentalCompile":false,"dynamicIds":[84888],"gssp":true,"scriptLoader":[]}