Frontiers in immunology · 2026
Use of explainable cluster analysis to identify distinct subtypes of Meige syndrome patients and associated biomarker profiles.
СтатьяНазвание и аннотация: Перевод в очереди
Обработка статьи · этапы, ошибки и управление
Загрузка состояния…
Ошибки и ограничения
История · было → стало
Состояние обновляется каждые 5 секунд, пока блок открыт. Процент показывается только там, где известен объём работы. Пауза останавливает продолжение этапа; текущий запрос к источнику может завершиться.
Sun Xiaoli, Feng Xinyu, Zhu Yingjie, Fu Runing
Аннотация
BACKGROUND: Meige syndrome (MS), a complex form of segmental craniocervical dystonia, currently lacks objective diagnostic biomarkers.This study aims to identify biological biomarkers that can stratify patients with MS and to reveal the complex pathophysiological heterogeneity of this disorder, with particular attention to the peripheral immune-metabolic profiles that may reflect underlying neuroimmune dysfunction.
METHODS: We retrospectively collected clinical and laboratory data from 1,782 patients with MS during their first hospitalization at the MS Center of The Third People's Hospital of Henan Province between 2023 and 2025. A total of 66 clinical and laboratory indicators were included. An interpretable clustering analysis framework was applied to perform a series of pipeline analyses, including imputation, dimensionality reduction, outlier removal, and clustering. Next, feature importance was assessed. A multi-class classification model was then constructed and evaluated using XGBoost. Finally, the Shapley Additive exPlanations(SHAP) algorithm was employed to interpret the contribution of each feature.
RESULTS: The preprocessing pipeline effectively imputed missing values (ranging from 0.06% to 14.14%) and removed outliers. Unsupervised clustering analysis revealed three heterogeneous subtypes of MS. A total of 23 important features were selected to construct a multi-class XGBoost classification model. The model demonstrated excellent performance, with a macro-average AUC of 0.9790 and an Obuchowski index of 0.9784. The SHAP analysis further elucidated the unique hematological and biochemical metabolic characteristics of each subtype.
CONCLUSION: This study establishes a novel, data-driven subtyping system for MS based on objective blood and metabolic profiles. The three identified subtypes display distinct immune-metabolic signatures: a hypercoagulable-inflammatory profile, a metabolic-excess profile with impaired vascular protection, and a frailty-associated antioxidant-deficient profile. These peripheral biomarker patterns suggest testable hypotheses regarding neuroimmune mechanisms-including fibrinogen-driven microglial activation, oxidized lipid-mediated blood-brain barrier disruption, and uric acid-dependent antioxidant depletion-that may contribute to the pathophysiology of different MS subtypes. This framework may inform future prospective investigations into precision medicine approaches for MS, though substantial validation is required before clinical translation.
METHODS: We retrospectively collected clinical and laboratory data from 1,782 patients with MS during their first hospitalization at the MS Center of The Third People's Hospital of Henan Province between 2023 and 2025. A total of 66 clinical and laboratory indicators were included. An interpretable clustering analysis framework was applied to perform a series of pipeline analyses, including imputation, dimensionality reduction, outlier removal, and clustering. Next, feature importance was assessed. A multi-class classification model was then constructed and evaluated using XGBoost. Finally, the Shapley Additive exPlanations(SHAP) algorithm was employed to interpret the contribution of each feature.
RESULTS: The preprocessing pipeline effectively imputed missing values (ranging from 0.06% to 14.14%) and removed outliers. Unsupervised clustering analysis revealed three heterogeneous subtypes of MS. A total of 23 important features were selected to construct a multi-class XGBoost classification model. The model demonstrated excellent performance, with a macro-average AUC of 0.9790 and an Obuchowski index of 0.9784. The SHAP analysis further elucidated the unique hematological and biochemical metabolic characteristics of each subtype.
CONCLUSION: This study establishes a novel, data-driven subtyping system for MS based on objective blood and metabolic profiles. The three identified subtypes display distinct immune-metabolic signatures: a hypercoagulable-inflammatory profile, a metabolic-excess profile with impaired vascular protection, and a frailty-associated antioxidant-deficient profile. These peripheral biomarker patterns suggest testable hypotheses regarding neuroimmune mechanisms-including fibrinogen-driven microglial activation, oxidized lipid-mediated blood-brain barrier disruption, and uric acid-dependent antioxidant depletion-that may contribute to the pathophysiology of different MS subtypes. This framework may inform future prospective investigations into precision medicine approaches for MS, though substantial validation is required before clinical translation.