Inflammation research : official journal of the European Histamine Research Society ... [et al.] · 2026
Machine learning and multi-omics integration identifies immunological predictors and mechanistic insights in autoimmune encephalitis.
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.
Guo Junshuang, Zou Yizhou
Аннотация
OBJECTIVE: To develop an interpretable prognostic prediction model for autoimmune encephalitis (AE) using immunological indicators and to investigate the potential role of nucleophosmin (NPM1) in disease pathogenesis through multi-omics approaches.
METHODS: We enrolled patients diagnosed with antibody-positive AE and analyzed a broad panel of immunological indicators. Prognostic prediction models were developed using eight machine learning algorithms and validated in an independent cohort. Model interpretability was enhanced through SHapley Additive exPlanations (SHAP) analysis. We further evaluated the therapeutic potential of protein A immunoadsorption (PAIA) in reducing pathogenic antibodies. Building upon these clinical and immunological findings, we sought to investigate the underlying mechanisms by exploring the role of nucleophosmin (NPM1). To this end, we integrated single-cell RNA sequencing and spatial transcriptomics in an experimental autoimmune encephalomyelitis (EAE) model and conducted a phenome-wide association study (PheWAS) to assess its safety as a potential therapeutic target candidate.
RESULTS: Six key immunological indicators were identified for model construction: cerebrospinal fluid /serum IgG quotient (QIgG), lymphocyte count, double negative T cell count, double positive T cell count, NK cell count, and T cell percentage. The RF, XGBoost, and LGBM models demonstrated high predictive performance, with AUC values of 0.978, 0.917, and 0.900, and accuracies of 0.940, 0.916, and 0.831, respectively. Anti-NMDAR antibody titers in cerebrospinal fluid decreased (from 1:3.2 to 1:1) following PAIA treatment in a single patient. Cell communication analysis revealed enhanced intercellular signaling in the high-Npm1 expression group, particularly involving the PSAP pathway. Spatial transcriptomics confirmed upregulated Npm1 expression in EAE lesions. PheWAS indicated no significant off-target effects associated with NPM1.
CONCLUSION: This study provides an interpretable prognostic framework for AE, presents preliminary evidence for PAIA, and nominates NPM1 as a potential mechanistic player in disease pathogenesis. Its suitability as a potential therapeutic target requires further safety validation, despite the absence of significant signals in the preliminary PheWAS.
METHODS: We enrolled patients diagnosed with antibody-positive AE and analyzed a broad panel of immunological indicators. Prognostic prediction models were developed using eight machine learning algorithms and validated in an independent cohort. Model interpretability was enhanced through SHapley Additive exPlanations (SHAP) analysis. We further evaluated the therapeutic potential of protein A immunoadsorption (PAIA) in reducing pathogenic antibodies. Building upon these clinical and immunological findings, we sought to investigate the underlying mechanisms by exploring the role of nucleophosmin (NPM1). To this end, we integrated single-cell RNA sequencing and spatial transcriptomics in an experimental autoimmune encephalomyelitis (EAE) model and conducted a phenome-wide association study (PheWAS) to assess its safety as a potential therapeutic target candidate.
RESULTS: Six key immunological indicators were identified for model construction: cerebrospinal fluid /serum IgG quotient (QIgG), lymphocyte count, double negative T cell count, double positive T cell count, NK cell count, and T cell percentage. The RF, XGBoost, and LGBM models demonstrated high predictive performance, with AUC values of 0.978, 0.917, and 0.900, and accuracies of 0.940, 0.916, and 0.831, respectively. Anti-NMDAR antibody titers in cerebrospinal fluid decreased (from 1:3.2 to 1:1) following PAIA treatment in a single patient. Cell communication analysis revealed enhanced intercellular signaling in the high-Npm1 expression group, particularly involving the PSAP pathway. Spatial transcriptomics confirmed upregulated Npm1 expression in EAE lesions. PheWAS indicated no significant off-target effects associated with NPM1.
CONCLUSION: This study provides an interpretable prognostic framework for AE, presents preliminary evidence for PAIA, and nominates NPM1 as a potential mechanistic player in disease pathogenesis. Its suitability as a potential therapeutic target requires further safety validation, despite the absence of significant signals in the preliminary PheWAS.