Development and validation of an age prediction model using mRNA markers.

As a downstream product of transcriptional regulation, mRNA offers valuable insights into age-associated molecular changes and shows considerable potential for applications in age estimation. In this study, RNA sequencing was conducted on peripheral blood samples from 127 healthy Chinese individuals spanning a broad age range (18-80 years). Differential expression analysis were performed on expression profiles among the young (＜30 years), middle-aged (30-60 years) and elderly (＞60 years) groups. A total of 79 differentially expressed genes (DEGs) were identified. Kyoto Encyclopedia of Genes and Genomes (KEGG) and Gene Ontology (GO) enrichment analysis revealed that the DEGs were mainly involved in response to type I interferon and cell-cell adhesion, which may reflect the activation of pro-inflammatory responses and neurodegenerative changes associated with aging. Through Spearman correlation analysis and the Lasso regression method, 34 candidate age-related genes (ARGs) were selected as predictive features. Notably, seven of these mRNAs (ARHGEF4, ARF6, AMIGO1, FITM2, PLEKHG4, SLC5A10, and HKR1) were identified for the first time as ARGs. Age prediction models were then constructed using five machine learning algorithms, with a 7:3 split for the train set and test set. Internal validation was conducted using 20 samples from the same cohort, while external validation was performed using two independent datasets: GSE262619 and GSE124326. Among them, the elastic net model showed the best performance, yielding a mean absolute error (MAE) of 6.72 years on the test set, 7.56 years on the internal validation set and 11.74 years on the external validation set, demonstrating strong robustness in age prediction. In general, this study delineated age-associated mRNA expression patterns and established a robust and accurate age prediction model based on a minimal set of 34 mRNA markers.