中国神经再生研究(英文版) ›› 2026, Vol. 21 ›› Issue (8): 3557-3558.doi: 10.4103/NRR.NRR-D-25-00940

• 观点:退行性病与再生 • 上一篇    下一篇

机器学习驱动的血液转录组分析对多发性硬化症的影响

  

  • 出版日期:2026-08-18 发布日期:2026-04-25

Impact of Machine Learning-Driven Analysis of Blood Transcriptomes in Multiple Sclerosis

Alessandro Digilio, Cinthia Farina*   

  1. Institute of Experimental Neurology and Division of Neuroscience, IRCCS San Raffaele Scientific Institute, Milan, Italy
  • Online:2026-08-18 Published:2026-04-25
  • Contact: Cinthia Farina, PhD, farina.cinthia@hsr.it.
  • Supported by:
    This work was supported by Italian Ministry for Health (RF-2011-02349698, RF-2018-12367731) (to CF).

摘要: https://orcid.org/0000-0002-4466-9676 (Cinthia Farina)

Abstract: Multiple sclerosis (MS) is a chronic disorder of the central nervous system characterized by multifocal lesions where inflammation, demyelination, and neurodegeneration occur (Jakimovski et al., 2024). MS diagnosis primarily relies on the demonstration of dissemination in time and space of the lesions based on clinical, magnetic resonance imaging (MRI), and cerebrospinal fluid assessments (Jakimovski et al., 2024). The disease can follow distinct clinical trajectories broadly described as relapsing-remitting MS (RR MS), the most common form characterized by acute episodes of neurological worsening followed by partial or complete recovery, and primary progressive MS (PP MS), where neurological disability accumulates steadily from onset (Jakimovski et al., 2024). After several years of disease, RR MS patients may also develop a progressive course (thus called secondary progressive MS, SP MS) (Jakimovski et al., 2024). It is well known that animal models resembling MS are mostly immune-mediated, that MS is accompanied by alterations in the immune system, and that approved drugs are immunomodulatory or immunosuppressive (Jakimovski et al., 2024). These observations underline the key role of peripheral immunity in the pathogenesis and maintenance of a neurological disorder and have prompted attention toward peripheral biomarkers to capture early systemic signals of disease. In this evolving landscape, blood transcriptomics has emerged as a valuable and minimally invasive tool to explore immune changes in MS systematically. While early studies identified blood-based transcriptomic signatures at distinct MS forms by classical differential gene expression analyses (Srinivasan et al., 2017a, b), the development of machine learning (ML) algorithms to model human disorders has since revolutionized our approach to biomarker definition. Still, while the application of ML to clinical and MRI data has shown great limits in predicting MS diagnosis and evolution (Bonacchi et al., 2022), it has provided excellent results when trained with blood transcriptomic data (Acquaviva et al., 2020; Omrani et al., 2024).