Scholarlink

<p>Aging is accompanied by profound alterations in the immune system; yet, an accurate prediction of immunological age remains challenging. While transcriptomic approaches have yielded insights into immune aging, protein-level profiling and machine learning-based prediction tools remain underdeveloped. Here, we employ mass cytometry to analyse murine splenic CD45⁺ immune cells across various age groups, profiling the expression of 30 protein markers and monitoring age-related immune changes. By analysing six major immune subsets (CD8⁺ T cells, CD4⁺ T cells, B cells, conventional type 1 and type 2 dendritic cells, and macrophages), we extract 103 molecular features and train a machine learning model using support vector regression (SVR) to predict immunological age. The model demonstrates robust generalizability by accurately predicting age in independent, test samples that were not used during model training. Furthermore, we confirm the robustness of our model using an obese mouse model, which exhibits metabolic dysfunction–associated immune senescence. Thus, our findings establish a robust framework for predicting immune-aging based on multidimensional protein expression data and machine learning. This tool enables quantitative assessment of immune aging and demonstrates strong translational potential for identifying obesity- and disease-related immune senescence.</p>

错误：搜索内容不能为空，请输入英文关键词
错误：关键词超出字数限制，请精简
高级检索

Reading the immune clock: a machine learning model predicts mouse immune age from cellular patterns

  • 期刊论文

  • 出版模式：

    开放获取

  • 发表日期：

    2025年12月10日

查看全文
  • Hyun Bo Sim,
  • Ji-Hun Jang,
  • Seul-Ki Mun,
  • Sonny C. Ramos,
  • Dae-Han Park,
  • Yu-Jeong Choi,
  • Ji Yeon Han,
  • Ju-Bin Lee,
  • Ho Seok Chung,
  • Kyung-Bok Lee,
  • Dong-Jo Chang,
  • Seung-Hyun Jeong,
  • Jong-Jin Kim

摘要

Aging is accompanied by profound alterations in the immune system; yet, an accurate prediction of immunological age remains challenging. While transcriptomic approaches have yielded insights into immune aging, protein-level profiling and machine learning-based prediction tools remain underdeveloped. Here, we employ mass cytometry to analyse murine splenic CD45⁺ immune cells across various age groups, profiling the expression of 30 protein markers and monitoring age-related immune changes. By analysing six major immune subsets (CD8⁺ T cells, CD4⁺ T cells, B cells, conventional type 1 and type 2 dendritic cells, and macrophages), we extract 103 molecular features and train a machine learning model using support vector regression (SVR) to predict immunological age. The model demonstrates robust generalizability by accurately predicting age in independent, test samples that were not used during model training. Furthermore, we confirm the robustness of our model using an obese mouse model, which exhibits metabolic dysfunction–associated immune senescence. Thus, our findings establish a robust framework for predicting immune-aging based on multidimensional protein expression data and machine learning. This tool enables quantitative assessment of immune aging and demonstrates strong translational potential for identifying obesity- and disease-related immune senescence.