New research published in
The team trained a proteomic aging clock using data from 55,319 participants in the UK Biobank, aged 50-85 years. The model demonstrated strong predictive accuracy overall, but its precision declined in the oldest group (ages 80-89). Here, the clock’s mean absolute error (MAE) was 4.08 years, reflecting greater biological variability at advanced ages. The model consistently overestimated age in younger participants (ages 50-59) by ~2.25 years but underestimated in the oldest group by ~4.07 years, reinforcing the idea that the complexity of aging processes intensifies with age.
To test biological relevance, the researchers applied the model to 84 COVID-19 patient samples with varying severity, due to their similar fibrotic lung sequelae. The clock revealed patients with severe cases, often associated with pulmonary fibrosis, showed significantly higher biological age compared to healthy controls (+2.77 years, P = .026).
In parallel, investigators developed ipf-P3GPT, a transformer-based generative AI model trained on datasets from human and in vitro studies of fibrotic disease. The model generated gene expression profiles for individuals with IPF and those experiencing normal lung aging (ages 30-70). It identified 96 genes linked to IPF and 93 genes associated with aging, with only 15 overlapping (15.6%), “suggesting substantial divergence in the underlying molecular programs.” More than half (53.3%) of these shared genes showed opposite regulation between IPF and aging.
“The differential regulation of key fibrotic pathways between IPF and aging generations suggests that IPF may represent dysregulation rather than mere acceleration of normal aging processes,” the authors wrote.
Both aging and IPF signatures demonstrated significant involvement of extracellular matrix (ECM)-associated genes, yet with distinct regulatory patterns. “This limited overlap provides critical insights into the relationship between aging and IPF pathogenesis,” the authors explain. Genes like COL15A1, ASPN, IL1RL1, IL6, PAPPA, and ADAMTS1 were identified as core overlapping processes involved in ECM remodeling, inflammatory signaling, and tissue repair. Some known drivers of fibrosis, such as MMP1, MMP13, AKT3, and IL-6, were uniquely present in the IPF signature, indicating that while aging increases vulnerability to IPF, the disease follows its own distinct pathological course, separate from the typical changes associated with aging.
The study acknowledges limitations, including reliance on computational models without direct experimental validation. While the COVID-19 dataset provided some indirect validation, it may not fully represent the pathology of IPF. Next steps include combining multi-omic data and testing results against clinical trial evidence, such as the phase 2 rentosertib trial in IPF.2
Idiopathic pulmonary fibrosis (IPF) is a chronic, progressive lung disease marked by excessive extracellular matrix buildup that drives loss of lung function and contributes to high mortality. It most often affects adults over the age of 60, underscoring a strong connection with aging. While available treatments can slow progression, lung transplantation remains the only intervention shown to extend survival. According to the authors, “Identifying the mechanisms shared by aging and fibrosis is crucial for developing targeted therapies that can potentially benefit the global population.” This methodology can potentially be extended to other fibrotic conditions.
References
1. Galkin F, Chen S, Aliper A, Zhavoronkov A, Ren F. AI-driven toolset for IPF and aging research associates lung fibrosis with accelerated aging. Aging. 2025;17(8):1999-2014. doi:10.18632/aging.206295
2. Xu, Z., Ren, F., Wang, P. et al. A generative AI-discovered TNIK inhibitor for idiopathic pulmonary fibrosis: a randomized phase 2a trial. Nat Med 2025;31:2602–2610. doi:10.1038/s41591-025-03743-2